PyObject *wrap_fit(PyObject *self,PyObject *args)
{
PyObject *cs1, *cs2, *mass;
if(!PyArg_ParseTuple(args,"OOO",&cs1, &cs2, &mass))
return NULL;
int natoms1 = PySequence_Length(cs1);
int natoms2 = PySequence_Length(cs2);
if( natoms1 != natoms2 ) {
Error("Cannot fit coordinate sets with different lengths");
}
rvec x1[natoms1];
rvec x2[natoms1];
real m[natoms1];
PyObject2rvec( cs1, x1, natoms1);
PyObject2rvec( cs2, x2, natoms2);
PyObject2real_array(mass, m, natoms1);
rvec cent;
center_and_get_vec(x1, natoms1, cent); // center x1 and get vector for back translation
center(x2, natoms1); // center x2
do_fit(natoms1, m, x1, x2);
int i;
for(i=0;i<natoms1;i++) // translate back
rvec_add( x2[i], cent, x2[i]);
PyObject *ret = rvec2PyObject(x2, natoms1);
return ret;
}
static void print_fitted_function(const char *fitfile,
const char *fn_fitted,
gmx_bool bXYdy,
int nset,
int n,
real *t,
real **val,
int npargs,
t_pargs *ppa,
gmx_output_env_t *oenv)
{
FILE *out_fit = gmx_ffopen(fitfile, "w");
if (bXYdy && nset >= 2)
{
do_fit(out_fit, 0, TRUE, n, t, val, npargs, ppa, oenv,
fn_fitted);
}
else
{
char *buf2 = NULL;
int s, buflen = 0;
if (NULL != fn_fitted)
{
buflen = std::strlen(fn_fitted)+32;
snew(buf2, buflen);
std::strncpy(buf2, fn_fitted, buflen);
buf2[std::strlen(buf2)-4] = '\0';
}
for (s = 0; s < nset; s++)
{
char *buf = NULL;
if (NULL != fn_fitted)
{
snew(buf, buflen);
snprintf(buf, n, "%s_%d.xvg", buf2, s);
}
do_fit(out_fit, s, FALSE, n, t, val, npargs, ppa, oenv, buf);
sfree(buf);
}
sfree(buf2);
}
gmx_ffclose(out_fit);
}
void FitACF(struct RadarParm *prm,
struct RawData *raw,struct FitBlock *input,
struct FitData *fit) {
int i,j;
int fnum,goose;
if (prm->time.yr < 1993) input->prm.old=1;
fit->revision.major=FITACF_MAJOR_REVISION;
fit->revision.minor=FITACF_MINOR_REVISION;
input->prm.xcf=prm->xcf;
input->prm.tfreq=prm->tfreq;
input->prm.noise=prm->noise.search;
input->prm.nrang=prm->nrang;
input->prm.smsep=prm->smsep;
input->prm.nave=prm->nave;
input->prm.mplgs=prm->mplgs;
input->prm.mpinc=prm->mpinc;
input->prm.txpl=prm->txpl;
input->prm.lagfr=prm->lagfr;
input->prm.mppul=prm->mppul;
input->prm.bmnum=prm->bmnum;
input->prm.cp=prm->cp;
input->prm.channel=prm->channel;
input->prm.offset=prm->offset; /* stereo offset */
/* need to incorporate Sessai's code for setting the offset
for legacy data here.
*/
for (i=0;i<input->prm.mppul;i++) input->prm.pulse[i]=prm->pulse[i];
for (i=0;i<input->prm.mplgs;i++) {
input->prm.lag[0][i]=prm->lag[i][0];
input->prm.lag[1][i]=prm->lag[i][1];
}
for (i=0;i<input->prm.nrang;i++) {
input->prm.pwr0[i]=raw->pwr0[i];
for (j=0;j<input->prm.mplgs;j++) {
input->acfd[i][j].x=raw->acfd[i][j][0];
input->acfd[i][j].y=raw->acfd[i][j][1];
}
if (input->prm.xcf) {
for (j=0;j<input->prm.mplgs;j++) {
input->xcfd[i][j].x=raw->xcfd[i][j][0];
input->xcfd[i][j].y=raw->xcfd[i][j][1];
}
}
}
goose=((prm->stid)==GOOSEBAY);
fnum=do_fit(input,5,goose,fit->rng,fit->xrng,fit->elv,&fit->noise);
}
int gmx_covar(int argc,char *argv[])
{
const char *desc[] = {
"[TT]g_covar[tt] calculates and diagonalizes the (mass-weighted)",
"covariance matrix.",
"All structures are fitted to the structure in the structure file.",
"When this is not a run input file periodicity will not be taken into",
"account. When the fit and analysis groups are identical and the analysis",
"is non mass-weighted, the fit will also be non mass-weighted.",
"[PAR]",
"The eigenvectors are written to a trajectory file ([TT]-v[tt]).",
"When the same atoms are used for the fit and the covariance analysis,",
"the reference structure for the fit is written first with t=-1.",
"The average (or reference when [TT]-ref[tt] is used) structure is",
"written with t=0, the eigenvectors",
"are written as frames with the eigenvector number as timestamp.",
"[PAR]",
"The eigenvectors can be analyzed with [TT]g_anaeig[tt].",
"[PAR]",
"Option [TT]-ascii[tt] writes the whole covariance matrix to",
"an ASCII file. The order of the elements is: x1x1, x1y1, x1z1, x1x2, ...",
"[PAR]",
"Option [TT]-xpm[tt] writes the whole covariance matrix to an [TT].xpm[tt] file.",
"[PAR]",
"Option [TT]-xpma[tt] writes the atomic covariance matrix to an [TT].xpm[tt] file,",
"i.e. for each atom pair the sum of the xx, yy and zz covariances is",
"written.",
"[PAR]",
"Note that the diagonalization of a matrix requires memory and time",
"that will increase at least as fast as than the square of the number",
"of atoms involved. It is easy to run out of memory, in which",
"case this tool will probably exit with a 'Segmentation fault'. You",
"should consider carefully whether a reduced set of atoms will meet",
"your needs for lower costs."
};
static gmx_bool bFit=TRUE,bRef=FALSE,bM=FALSE,bPBC=TRUE;
static int end=-1;
t_pargs pa[] = {
{ "-fit", FALSE, etBOOL, {&bFit},
"Fit to a reference structure"},
{ "-ref", FALSE, etBOOL, {&bRef},
"Use the deviation from the conformation in the structure file instead of from the average" },
{ "-mwa", FALSE, etBOOL, {&bM},
"Mass-weighted covariance analysis"},
{ "-last", FALSE, etINT, {&end},
"Last eigenvector to write away (-1 is till the last)" },
{ "-pbc", FALSE, etBOOL, {&bPBC},
"Apply corrections for periodic boundary conditions" }
};
FILE *out;
t_trxstatus *status;
t_trxstatus *trjout;
t_topology top;
int ePBC;
t_atoms *atoms;
rvec *x,*xread,*xref,*xav,*xproj;
matrix box,zerobox;
real *sqrtm,*mat,*eigval,sum,trace,inv_nframes;
real t,tstart,tend,**mat2;
real xj,*w_rls=NULL;
real min,max,*axis;
int ntopatoms,step;
int natoms,nat,count,nframes0,nframes,nlevels;
gmx_large_int_t ndim,i,j,k,l;
int WriteXref;
const char *fitfile,*trxfile,*ndxfile;
const char *eigvalfile,*eigvecfile,*averfile,*logfile;
const char *asciifile,*xpmfile,*xpmafile;
char str[STRLEN],*fitname,*ananame,*pcwd;
int d,dj,nfit;
atom_id *index,*ifit;
gmx_bool bDiffMass1,bDiffMass2;
time_t now;
char timebuf[STRLEN];
t_rgb rlo,rmi,rhi;
real *tmp;
output_env_t oenv;
gmx_rmpbc_t gpbc=NULL;
t_filenm fnm[] = {
{ efTRX, "-f", NULL, ffREAD },
{ efTPS, NULL, NULL, ffREAD },
{ efNDX, NULL, NULL, ffOPTRD },
{ efXVG, NULL, "eigenval", ffWRITE },
{ efTRN, "-v", "eigenvec", ffWRITE },
{ efSTO, "-av", "average.pdb", ffWRITE },
{ efLOG, NULL, "covar", ffWRITE },
{ efDAT, "-ascii","covar", ffOPTWR },
{ efXPM, "-xpm","covar", ffOPTWR },
{ efXPM, "-xpma","covara", ffOPTWR }
};
#define NFILE asize(fnm)
CopyRight(stderr,argv[0]);
parse_common_args(&argc,argv,PCA_CAN_TIME | PCA_TIME_UNIT | PCA_BE_NICE,
NFILE,fnm,asize(pa),pa,asize(desc),desc,0,NULL,&oenv);
clear_mat(zerobox);
fitfile = ftp2fn(efTPS,NFILE,fnm);
trxfile = ftp2fn(efTRX,NFILE,fnm);
ndxfile = ftp2fn_null(efNDX,NFILE,fnm);
eigvalfile = ftp2fn(efXVG,NFILE,fnm);
eigvecfile = ftp2fn(efTRN,NFILE,fnm);
averfile = ftp2fn(efSTO,NFILE,fnm);
logfile = ftp2fn(efLOG,NFILE,fnm);
asciifile = opt2fn_null("-ascii",NFILE,fnm);
xpmfile = opt2fn_null("-xpm",NFILE,fnm);
xpmafile = opt2fn_null("-xpma",NFILE,fnm);
read_tps_conf(fitfile,str,&top,&ePBC,&xref,NULL,box,TRUE);
atoms=&top.atoms;
if (bFit) {
printf("\nChoose a group for the least squares fit\n");
get_index(atoms,ndxfile,1,&nfit,&ifit,&fitname);
if (nfit < 3)
gmx_fatal(FARGS,"Need >= 3 points to fit!\n");
} else
nfit=0;
printf("\nChoose a group for the covariance analysis\n");
get_index(atoms,ndxfile,1,&natoms,&index,&ananame);
bDiffMass1=FALSE;
if (bFit) {
snew(w_rls,atoms->nr);
for(i=0; (i<nfit); i++) {
w_rls[ifit[i]]=atoms->atom[ifit[i]].m;
if (i)
bDiffMass1 = bDiffMass1 || (w_rls[ifit[i]]!=w_rls[ifit[i-1]]);
}
}
bDiffMass2=FALSE;
snew(sqrtm,natoms);
for(i=0; (i<natoms); i++)
if (bM) {
sqrtm[i]=sqrt(atoms->atom[index[i]].m);
if (i)
bDiffMass2 = bDiffMass2 || (sqrtm[i]!=sqrtm[i-1]);
}
else
sqrtm[i]=1.0;
if (bFit && bDiffMass1 && !bDiffMass2) {
bDiffMass1 = natoms != nfit;
i=0;
for (i=0; (i<natoms) && !bDiffMass1; i++)
bDiffMass1 = index[i] != ifit[i];
if (!bDiffMass1) {
fprintf(stderr,"\n"
"Note: the fit and analysis group are identical,\n"
" while the fit is mass weighted and the analysis is not.\n"
" Making the fit non mass weighted.\n\n");
for(i=0; (i<nfit); i++)
w_rls[ifit[i]]=1.0;
}
}
/* Prepare reference frame */
if (bPBC) {
gpbc = gmx_rmpbc_init(&top.idef,ePBC,atoms->nr,box);
gmx_rmpbc(gpbc,atoms->nr,box,xref);
}
if (bFit)
reset_x(nfit,ifit,atoms->nr,NULL,xref,w_rls);
snew(x,natoms);
snew(xav,natoms);
ndim=natoms*DIM;
if (sqrt(GMX_LARGE_INT_MAX)<ndim) {
gmx_fatal(FARGS,"Number of degrees of freedoms to large for matrix.\n");
}
snew(mat,ndim*ndim);
fprintf(stderr,"Calculating the average structure ...\n");
nframes0 = 0;
nat=read_first_x(oenv,&status,trxfile,&t,&xread,box);
if (nat != atoms->nr)
fprintf(stderr,"\nWARNING: number of atoms in tpx (%d) and trajectory (%d) do not match\n",natoms,nat);
do {
nframes0++;
/* calculate x: a fitted struture of the selected atoms */
if (bPBC)
gmx_rmpbc(gpbc,nat,box,xread);
if (bFit) {
reset_x(nfit,ifit,nat,NULL,xread,w_rls);
do_fit(nat,w_rls,xref,xread);
}
for (i=0; i<natoms; i++)
rvec_inc(xav[i],xread[index[i]]);
} while (read_next_x(oenv,status,&t,nat,xread,box));
close_trj(status);
inv_nframes = 1.0/nframes0;
for(i=0; i<natoms; i++)
for(d=0; d<DIM; d++) {
xav[i][d] *= inv_nframes;
xread[index[i]][d] = xav[i][d];
}
write_sto_conf_indexed(opt2fn("-av",NFILE,fnm),"Average structure",
atoms,xread,NULL,epbcNONE,zerobox,natoms,index);
sfree(xread);
fprintf(stderr,"Constructing covariance matrix (%dx%d) ...\n",(int)ndim,(int)ndim);
nframes=0;
nat=read_first_x(oenv,&status,trxfile,&t,&xread,box);
tstart = t;
do {
nframes++;
tend = t;
/* calculate x: a (fitted) structure of the selected atoms */
if (bPBC)
gmx_rmpbc(gpbc,nat,box,xread);
if (bFit) {
reset_x(nfit,ifit,nat,NULL,xread,w_rls);
do_fit(nat,w_rls,xref,xread);
}
if (bRef)
for (i=0; i<natoms; i++)
rvec_sub(xread[index[i]],xref[index[i]],x[i]);
else
for (i=0; i<natoms; i++)
rvec_sub(xread[index[i]],xav[i],x[i]);
for (j=0; j<natoms; j++) {
for (dj=0; dj<DIM; dj++) {
k=ndim*(DIM*j+dj);
xj=x[j][dj];
for (i=j; i<natoms; i++) {
l=k+DIM*i;
for(d=0; d<DIM; d++)
mat[l+d] += x[i][d]*xj;
}
}
}
} while (read_next_x(oenv,status,&t,nat,xread,box) &&
(bRef || nframes < nframes0));
close_trj(status);
gmx_rmpbc_done(gpbc);
fprintf(stderr,"Read %d frames\n",nframes);
if (bRef) {
/* copy the reference structure to the ouput array x */
snew(xproj,natoms);
for (i=0; i<natoms; i++)
copy_rvec(xref[index[i]],xproj[i]);
} else {
xproj = xav;
}
/* correct the covariance matrix for the mass */
inv_nframes = 1.0/nframes;
for (j=0; j<natoms; j++)
for (dj=0; dj<DIM; dj++)
for (i=j; i<natoms; i++) {
k = ndim*(DIM*j+dj)+DIM*i;
for (d=0; d<DIM; d++)
mat[k+d] = mat[k+d]*inv_nframes*sqrtm[i]*sqrtm[j];
}
/* symmetrize the matrix */
for (j=0; j<ndim; j++)
for (i=j; i<ndim; i++)
mat[ndim*i+j]=mat[ndim*j+i];
trace=0;
for(i=0; i<ndim; i++)
trace+=mat[i*ndim+i];
fprintf(stderr,"\nTrace of the covariance matrix: %g (%snm^2)\n",
trace,bM ? "u " : "");
if (asciifile) {
out = ffopen(asciifile,"w");
for (j=0; j<ndim; j++) {
for (i=0; i<ndim; i+=3)
fprintf(out,"%g %g %g\n",
mat[ndim*j+i],mat[ndim*j+i+1],mat[ndim*j+i+2]);
}
ffclose(out);
}
if (xpmfile) {
min = 0;
max = 0;
snew(mat2,ndim);
for (j=0; j<ndim; j++) {
mat2[j] = &(mat[ndim*j]);
for (i=0; i<=j; i++) {
if (mat2[j][i] < min)
min = mat2[j][i];
if (mat2[j][j] > max)
max = mat2[j][i];
}
}
snew(axis,ndim);
for(i=0; i<ndim; i++)
axis[i] = i+1;
rlo.r = 0; rlo.g = 0; rlo.b = 1;
rmi.r = 1; rmi.g = 1; rmi.b = 1;
rhi.r = 1; rhi.g = 0; rhi.b = 0;
out = ffopen(xpmfile,"w");
nlevels = 80;
write_xpm3(out,0,"Covariance",bM ? "u nm^2" : "nm^2",
"dim","dim",ndim,ndim,axis,axis,
mat2,min,0.0,max,rlo,rmi,rhi,&nlevels);
ffclose(out);
sfree(axis);
sfree(mat2);
}
if (xpmafile) {
min = 0;
max = 0;
snew(mat2,ndim/DIM);
for (i=0; i<ndim/DIM; i++)
snew(mat2[i],ndim/DIM);
for (j=0; j<ndim/DIM; j++) {
for (i=0; i<=j; i++) {
mat2[j][i] = 0;
for(d=0; d<DIM; d++)
mat2[j][i] += mat[ndim*(DIM*j+d)+DIM*i+d];
if (mat2[j][i] < min)
min = mat2[j][i];
if (mat2[j][j] > max)
max = mat2[j][i];
mat2[i][j] = mat2[j][i];
}
}
snew(axis,ndim/DIM);
for(i=0; i<ndim/DIM; i++)
axis[i] = i+1;
rlo.r = 0; rlo.g = 0; rlo.b = 1;
rmi.r = 1; rmi.g = 1; rmi.b = 1;
rhi.r = 1; rhi.g = 0; rhi.b = 0;
out = ffopen(xpmafile,"w");
nlevels = 80;
write_xpm3(out,0,"Covariance",bM ? "u nm^2" : "nm^2",
"atom","atom",ndim/DIM,ndim/DIM,axis,axis,
mat2,min,0.0,max,rlo,rmi,rhi,&nlevels);
ffclose(out);
sfree(axis);
for (i=0; i<ndim/DIM; i++)
sfree(mat2[i]);
sfree(mat2);
}
/* call diagonalization routine */
fprintf(stderr,"\nDiagonalizing ...\n");
fflush(stderr);
snew(eigval,ndim);
snew(tmp,ndim*ndim);
memcpy(tmp,mat,ndim*ndim*sizeof(real));
eigensolver(tmp,ndim,0,ndim,eigval,mat);
sfree(tmp);
/* now write the output */
sum=0;
for(i=0; i<ndim; i++)
sum+=eigval[i];
fprintf(stderr,"\nSum of the eigenvalues: %g (%snm^2)\n",
sum,bM ? "u " : "");
if (fabs(trace-sum)>0.01*trace)
fprintf(stderr,"\nWARNING: eigenvalue sum deviates from the trace of the covariance matrix\n");
fprintf(stderr,"\nWriting eigenvalues to %s\n",eigvalfile);
sprintf(str,"(%snm\\S2\\N)",bM ? "u " : "");
out=xvgropen(eigvalfile,
"Eigenvalues of the covariance matrix",
"Eigenvector index",str,oenv);
for (i=0; (i<ndim); i++)
fprintf (out,"%10d %g\n",(int)i+1,eigval[ndim-1-i]);
ffclose(out);
if (end==-1) {
if (nframes-1 < ndim)
end=nframes-1;
else
end=ndim;
}
if (bFit) {
/* misuse lambda: 0/1 mass weighted analysis no/yes */
if (nfit==natoms) {
WriteXref = eWXR_YES;
for(i=0; i<nfit; i++)
copy_rvec(xref[ifit[i]],x[i]);
} else
WriteXref = eWXR_NO;
} else {
/* misuse lambda: -1 for no fit */
WriteXref = eWXR_NOFIT;
}
write_eigenvectors(eigvecfile,natoms,mat,TRUE,1,end,
WriteXref,x,bDiffMass1,xproj,bM,eigval);
out = ffopen(logfile,"w");
time(&now);
gmx_ctime_r(&now,timebuf,STRLEN);
fprintf(out,"Covariance analysis log, written %s\n",timebuf);
fprintf(out,"Program: %s\n",argv[0]);
#if ((defined WIN32 || defined _WIN32 || defined WIN64 || defined _WIN64) && !defined __CYGWIN__ && !defined __CYGWIN32__)
pcwd=_getcwd(str,STRLEN);
#else
pcwd=getcwd(str,STRLEN);
#endif
if(NULL==pcwd)
{
gmx_fatal(FARGS,"Current working directory is undefined");
}
fprintf(out,"Working directory: %s\n\n",str);
fprintf(out,"Read %d frames from %s (time %g to %g %s)\n",nframes,trxfile,
output_env_conv_time(oenv,tstart),output_env_conv_time(oenv,tend),output_env_get_time_unit(oenv));
if (bFit)
fprintf(out,"Read reference structure for fit from %s\n",fitfile);
if (ndxfile)
fprintf(out,"Read index groups from %s\n",ndxfile);
fprintf(out,"\n");
fprintf(out,"Analysis group is '%s' (%d atoms)\n",ananame,natoms);
if (bFit)
fprintf(out,"Fit group is '%s' (%d atoms)\n",fitname,nfit);
else
fprintf(out,"No fit was used\n");
fprintf(out,"Analysis is %smass weighted\n", bDiffMass2 ? "":"non-");
if (bFit)
fprintf(out,"Fit is %smass weighted\n", bDiffMass1 ? "":"non-");
fprintf(out,"Diagonalized the %dx%d covariance matrix\n",(int)ndim,(int)ndim);
fprintf(out,"Trace of the covariance matrix before diagonalizing: %g\n",
trace);
fprintf(out,"Trace of the covariance matrix after diagonalizing: %g\n\n",
sum);
fprintf(out,"Wrote %d eigenvalues to %s\n",(int)ndim,eigvalfile);
if (WriteXref == eWXR_YES)
fprintf(out,"Wrote reference structure to %s\n",eigvecfile);
fprintf(out,"Wrote average structure to %s and %s\n",averfile,eigvecfile);
fprintf(out,"Wrote eigenvectors %d to %d to %s\n",1,end,eigvecfile);
ffclose(out);
fprintf(stderr,"Wrote the log to %s\n",logfile);
thanx(stderr);
return 0;
}
static void project(const char *trajfile,t_topology *top,int ePBC,matrix topbox,
const char *projfile,const char *twodplotfile,
const char *threedplotfile, const char *filterfile,int skip,
const char *extremefile,gmx_bool bExtrAll,real extreme,
int nextr, t_atoms *atoms,int natoms,atom_id *index,
gmx_bool bFit,rvec *xref,int nfit,atom_id *ifit,real *w_rls,
real *sqrtm,rvec *xav,
int *eignr,rvec **eigvec,
int noutvec,int *outvec, gmx_bool bSplit,
const output_env_t oenv)
{
FILE *xvgrout=NULL;
int nat,i,j,d,v,vec,nfr,nframes=0,snew_size,frame;
t_trxstatus *out=NULL;
t_trxstatus *status;
int noutvec_extr,imin,imax;
real *pmin,*pmax;
atom_id *all_at;
matrix box;
rvec *xread,*x;
real t,inp,**inprod=NULL,min=0,max=0;
char str[STRLEN],str2[STRLEN],**ylabel,*c;
real fact;
gmx_rmpbc_t gpbc=NULL;
snew(x,natoms);
if (bExtrAll)
noutvec_extr=noutvec;
else
noutvec_extr=1;
if (trajfile) {
snew(inprod,noutvec+1);
if (filterfile) {
fprintf(stderr,"Writing a filtered trajectory to %s using eigenvectors\n",
filterfile);
for(i=0; i<noutvec; i++)
fprintf(stderr,"%d ",outvec[i]+1);
fprintf(stderr,"\n");
out=open_trx(filterfile,"w");
}
snew_size=0;
nfr=0;
nframes=0;
nat=read_first_x(oenv,&status,trajfile,&t,&xread,box);
if (nat>atoms->nr)
gmx_fatal(FARGS,"the number of atoms in your trajectory (%d) is larger than the number of atoms in your structure file (%d)",nat,atoms->nr);
snew(all_at,nat);
if (top)
gpbc = gmx_rmpbc_init(&top->idef,ePBC,nat,box);
for(i=0; i<nat; i++)
all_at[i]=i;
do {
if (nfr % skip == 0) {
if (top)
gmx_rmpbc(gpbc,nat,box,xread);
if (nframes>=snew_size) {
snew_size+=100;
for(i=0; i<noutvec+1; i++)
srenew(inprod[i],snew_size);
}
inprod[noutvec][nframes]=t;
/* calculate x: a fitted struture of the selected atoms */
if (bFit) {
reset_x(nfit,ifit,nat,NULL,xread,w_rls);
do_fit(nat,w_rls,xref,xread);
}
for (i=0; i<natoms; i++)
copy_rvec(xread[index[i]],x[i]);
for(v=0; v<noutvec; v++) {
vec=outvec[v];
/* calculate (mass-weighted) projection */
inp=0;
for (i=0; i<natoms; i++) {
inp+=(eigvec[vec][i][0]*(x[i][0]-xav[i][0])+
eigvec[vec][i][1]*(x[i][1]-xav[i][1])+
eigvec[vec][i][2]*(x[i][2]-xav[i][2]))*sqrtm[i];
}
inprod[v][nframes]=inp;
}
if (filterfile) {
for(i=0; i<natoms; i++)
for(d=0; d<DIM; d++) {
/* misuse xread for output */
xread[index[i]][d] = xav[i][d];
for(v=0; v<noutvec; v++)
xread[index[i]][d] +=
inprod[v][nframes]*eigvec[outvec[v]][i][d]/sqrtm[i];
}
write_trx(out,natoms,index,atoms,0,t,box,xread,NULL,NULL);
}
nframes++;
}
nfr++;
} while (read_next_x(oenv,status,&t,nat,xread,box));
close_trx(status);
sfree(x);
if (filterfile)
close_trx(out);
}
else
snew(xread,atoms->nr);
if (top)
gmx_rmpbc_done(gpbc);
if (projfile) {
snew(ylabel,noutvec);
for(v=0; v<noutvec; v++) {
sprintf(str,"vec %d",eignr[outvec[v]]+1);
ylabel[v]=strdup(str);
}
sprintf(str,"projection on eigenvectors (%s)",proj_unit);
write_xvgr_graphs(projfile, noutvec, 1, str, NULL, output_env_get_xvgr_tlabel(oenv),
(const char **)ylabel,
nframes, inprod[noutvec], inprod, NULL,
output_env_get_time_factor(oenv), FALSE, bSplit,oenv);
}
if (twodplotfile) {
sprintf(str,"projection on eigenvector %d (%s)",
eignr[outvec[0]]+1,proj_unit);
sprintf(str2,"projection on eigenvector %d (%s)",
eignr[outvec[noutvec-1]]+1,proj_unit);
xvgrout=xvgropen(twodplotfile,"2D projection of trajectory",str,str2,
oenv);
for(i=0; i<nframes; i++) {
if ( bSplit && i>0 && abs(inprod[noutvec][i])<1e-5 )
fprintf(xvgrout,"&\n");
fprintf(xvgrout,"%10.5f %10.5f\n",inprod[0][i],inprod[noutvec-1][i]);
}
ffclose(xvgrout);
}
if (threedplotfile) {
t_atoms atoms;
rvec *x;
real *b=NULL;
matrix box;
char *resnm,*atnm, pdbform[STRLEN];
gmx_bool bPDB, b4D;
FILE *out;
if (noutvec < 3)
gmx_fatal(FARGS,"You have selected less than 3 eigenvectors");
/* initialize */
bPDB = fn2ftp(threedplotfile)==efPDB;
clear_mat(box);
box[XX][XX] = box[YY][YY] = box[ZZ][ZZ] = 1;
b4D = bPDB && (noutvec >= 4);
if (b4D) {
fprintf(stderr, "You have selected four or more eigenvectors:\n"
"fourth eigenvector will be plotted "
"in bfactor field of pdb file\n");
sprintf(str,"4D proj. of traj. on eigenv. %d, %d, %d and %d",
eignr[outvec[0]]+1,eignr[outvec[1]]+1,
eignr[outvec[2]]+1,eignr[outvec[3]]+1);
} else {
sprintf(str,"3D proj. of traj. on eigenv. %d, %d and %d",
eignr[outvec[0]]+1,eignr[outvec[1]]+1,eignr[outvec[2]]+1);
}
init_t_atoms(&atoms,nframes,FALSE);
snew(x,nframes);
snew(b,nframes);
atnm=strdup("C");
resnm=strdup("PRJ");
if(nframes>10000)
fact=10000.0/nframes;
else
fact=1.0;
for(i=0; i<nframes; i++) {
atoms.atomname[i] = &atnm;
atoms.atom[i].resind = i;
atoms.resinfo[i].name = &resnm;
atoms.resinfo[i].nr = ceil(i*fact);
atoms.resinfo[i].ic = ' ';
x[i][XX]=inprod[0][i];
x[i][YY]=inprod[1][i];
x[i][ZZ]=inprod[2][i];
if (b4D)
b[i] =inprod[3][i];
}
if ( ( b4D || bSplit ) && bPDB ) {
strcpy(pdbform,get_pdbformat());
strcat(pdbform,"%8.4f%8.4f\n");
out=ffopen(threedplotfile,"w");
fprintf(out,"HEADER %s\n",str);
if ( b4D )
fprintf(out,"REMARK %s\n","fourth dimension plotted as B-factor");
j=0;
for(i=0; i<atoms.nr; i++) {
if ( j>0 && bSplit && abs(inprod[noutvec][i])<1e-5 ) {
fprintf(out,"TER\n");
j=0;
}
fprintf(out,pdbform,"ATOM",i+1,"C","PRJ",' ',j+1,
PR_VEC(10*x[i]), 1.0, 10*b[i]);
if (j>0)
fprintf(out,"CONECT%5d%5d\n", i, i+1);
j++;
}
fprintf(out,"TER\n");
ffclose(out);
} else
write_sto_conf(threedplotfile,str,&atoms,x,NULL,ePBC,box);
free_t_atoms(&atoms,FALSE);
}
if (extremefile) {
snew(pmin,noutvec_extr);
snew(pmax,noutvec_extr);
if (extreme==0) {
fprintf(stderr,"%11s %17s %17s\n","eigenvector","Minimum","Maximum");
fprintf(stderr,
"%11s %10s %10s %10s %10s\n","","value","frame","value","frame");
imin = 0;
imax = 0;
for(v=0; v<noutvec_extr; v++) {
for(i=0; i<nframes; i++) {
if (inprod[v][i]<inprod[v][imin])
imin = i;
if (inprod[v][i]>inprod[v][imax])
imax = i;
}
pmin[v] = inprod[v][imin];
pmax[v] = inprod[v][imax];
fprintf(stderr,"%7d %10.6f %10d %10.6f %10d\n",
eignr[outvec[v]]+1,
pmin[v],imin,pmax[v],imax);
}
}
else {
pmin[0] = -extreme;
pmax[0] = extreme;
}
/* build format string for filename: */
strcpy(str,extremefile);/* copy filename */
c=strrchr(str,'.'); /* find where extention begins */
strcpy(str2,c); /* get extention */
sprintf(c,"%%d%s",str2); /* append '%s' and extention to filename */
for(v=0; v<noutvec_extr; v++) {
/* make filename using format string */
if (noutvec_extr==1)
strcpy(str2,extremefile);
else
sprintf(str2,str,eignr[outvec[v]]+1);
fprintf(stderr,"Writing %d frames along eigenvector %d to %s\n",
nextr,outvec[v]+1,str2);
out=open_trx(str2,"w");
for(frame=0; frame<nextr; frame++) {
if ((extreme==0) && (nextr<=3))
for(i=0; i<natoms; i++) {
atoms->resinfo[atoms->atom[index[i]].resind].chainid = 'A' + frame;
}
for(i=0; i<natoms; i++)
for(d=0; d<DIM; d++)
xread[index[i]][d] =
(xav[i][d] + (pmin[v]*(nextr-frame-1)+pmax[v]*frame)/(nextr-1)
*eigvec[outvec[v]][i][d]/sqrtm[i]);
write_trx(out,natoms,index,atoms,0,frame,topbox,xread,NULL,NULL);
}
close_trx(out);
}
sfree(pmin);
sfree(pmax);
}
fprintf(stderr,"\n");
}
int gmx_analyze(int argc, char *argv[])
{
static const char *desc[] = {
"[TT]g_analyze[tt] reads an ASCII file and analyzes data sets.",
"A line in the input file may start with a time",
"(see option [TT]-time[tt]) and any number of [IT]y[it]-values may follow.",
"Multiple sets can also be",
"read when they are separated by & (option [TT]-n[tt]);",
"in this case only one [IT]y[it]-value is read from each line.",
"All lines starting with # and @ are skipped.",
"All analyses can also be done for the derivative of a set",
"(option [TT]-d[tt]).[PAR]",
"All options, except for [TT]-av[tt] and [TT]-power[tt], assume that the",
"points are equidistant in time.[PAR]",
"[TT]g_analyze[tt] always shows the average and standard deviation of each",
"set, as well as the relative deviation of the third",
"and fourth cumulant from those of a Gaussian distribution with the same",
"standard deviation.[PAR]",
"Option [TT]-ac[tt] produces the autocorrelation function(s).",
"Be sure that the time interval between data points is",
"much shorter than the time scale of the autocorrelation.[PAR]",
"Option [TT]-cc[tt] plots the resemblance of set i with a cosine of",
"i/2 periods. The formula is:[BR]"
"[MATH]2 ([INT][FROM]0[from][TO]T[to][int] y(t) [COS]i [GRK]pi[grk] t[cos] dt)^2 / [INT][FROM]0[from][TO]T[to][int] y^2(t) dt[math][BR]",
"This is useful for principal components obtained from covariance",
"analysis, since the principal components of random diffusion are",
"pure cosines.[PAR]",
"Option [TT]-msd[tt] produces the mean square displacement(s).[PAR]",
"Option [TT]-dist[tt] produces distribution plot(s).[PAR]",
"Option [TT]-av[tt] produces the average over the sets.",
"Error bars can be added with the option [TT]-errbar[tt].",
"The errorbars can represent the standard deviation, the error",
"(assuming the points are independent) or the interval containing",
"90% of the points, by discarding 5% of the points at the top and",
"the bottom.[PAR]",
"Option [TT]-ee[tt] produces error estimates using block averaging.",
"A set is divided in a number of blocks and averages are calculated for",
"each block. The error for the total average is calculated from",
"the variance between averages of the m blocks B[SUB]i[sub] as follows:",
"error^2 = [SUM][sum] (B[SUB]i[sub] - [CHEVRON]B[chevron])^2 / (m*(m-1)).",
"These errors are plotted as a function of the block size.",
"Also an analytical block average curve is plotted, assuming",
"that the autocorrelation is a sum of two exponentials.",
"The analytical curve for the block average is:[BR]",
"[MATH]f(t) = [GRK]sigma[grk][TT]*[tt][SQRT]2/T ( [GRK]alpha[grk] ([GRK]tau[grk][SUB]1[sub] (([EXP]-t/[GRK]tau[grk][SUB]1[sub][exp] - 1) [GRK]tau[grk][SUB]1[sub]/t + 1)) +[BR]",
" (1-[GRK]alpha[grk]) ([GRK]tau[grk][SUB]2[sub] (([EXP]-t/[GRK]tau[grk][SUB]2[sub][exp] - 1) [GRK]tau[grk][SUB]2[sub]/t + 1)))[sqrt][math],[BR]"
"where T is the total time.",
"[GRK]alpha[grk], [GRK]tau[grk][SUB]1[sub] and [GRK]tau[grk][SUB]2[sub] are obtained by fitting f^2(t) to error^2.",
"When the actual block average is very close to the analytical curve,",
"the error is [MATH][GRK]sigma[grk][TT]*[tt][SQRT]2/T (a [GRK]tau[grk][SUB]1[sub] + (1-a) [GRK]tau[grk][SUB]2[sub])[sqrt][math].",
"The complete derivation is given in",
"B. Hess, J. Chem. Phys. 116:209-217, 2002.[PAR]",
"Option [TT]-bal[tt] finds and subtracts the ultrafast \"ballistic\"",
"component from a hydrogen bond autocorrelation function by the fitting",
"of a sum of exponentials, as described in e.g.",
"O. Markovitch, J. Chem. Phys. 129:084505, 2008. The fastest term",
"is the one with the most negative coefficient in the exponential,",
"or with [TT]-d[tt], the one with most negative time derivative at time 0.",
"[TT]-nbalexp[tt] sets the number of exponentials to fit.[PAR]",
"Option [TT]-gem[tt] fits bimolecular rate constants ka and kb",
"(and optionally kD) to the hydrogen bond autocorrelation function",
"according to the reversible geminate recombination model. Removal of",
"the ballistic component first is strongly advised. The model is presented in",
"O. Markovitch, J. Chem. Phys. 129:084505, 2008.[PAR]",
"Option [TT]-filter[tt] prints the RMS high-frequency fluctuation",
"of each set and over all sets with respect to a filtered average.",
"The filter is proportional to cos([GRK]pi[grk] t/len) where t goes from -len/2",
"to len/2. len is supplied with the option [TT]-filter[tt].",
"This filter reduces oscillations with period len/2 and len by a factor",
"of 0.79 and 0.33 respectively.[PAR]",
"Option [TT]-g[tt] fits the data to the function given with option",
"[TT]-fitfn[tt].[PAR]",
"Option [TT]-power[tt] fits the data to [MATH]b t^a[math], which is accomplished",
"by fitting to [MATH]a t + b[math] on log-log scale. All points after the first",
"zero or with a negative value are ignored.[PAR]"
"Option [TT]-luzar[tt] performs a Luzar & Chandler kinetics analysis",
"on output from [TT]g_hbond[tt]. The input file can be taken directly",
"from [TT]g_hbond -ac[tt], and then the same result should be produced."
};
static real tb = -1, te = -1, frac = 0.5, filtlen = 0, binwidth = 0.1, aver_start = 0;
static gmx_bool bHaveT = TRUE, bDer = FALSE, bSubAv = TRUE, bAverCorr = FALSE, bXYdy = FALSE;
static gmx_bool bEESEF = FALSE, bEENLC = FALSE, bEeFitAc = FALSE, bPower = FALSE;
static gmx_bool bIntegrate = FALSE, bRegression = FALSE, bLuzar = FALSE, bLuzarError = FALSE;
static int nsets_in = 1, d = 1, nb_min = 4, resol = 10, nBalExp = 4, nFitPoints = 100;
static real temp = 298.15, fit_start = 1, fit_end = 60, smooth_tail_start = -1, balTime = 0.2, diffusion = 5e-5, rcut = 0.35;
/* must correspond to enum avbar* declared at beginning of file */
static const char *avbar_opt[avbarNR+1] = {
NULL, "none", "stddev", "error", "90", NULL
};
t_pargs pa[] = {
{ "-time", FALSE, etBOOL, {&bHaveT},
"Expect a time in the input" },
{ "-b", FALSE, etREAL, {&tb},
"First time to read from set" },
{ "-e", FALSE, etREAL, {&te},
"Last time to read from set" },
{ "-n", FALSE, etINT, {&nsets_in},
"Read this number of sets separated by &" },
{ "-d", FALSE, etBOOL, {&bDer},
"Use the derivative" },
{ "-dp", FALSE, etINT, {&d},
"HIDDENThe derivative is the difference over this number of points" },
{ "-bw", FALSE, etREAL, {&binwidth},
"Binwidth for the distribution" },
{ "-errbar", FALSE, etENUM, {avbar_opt},
"Error bars for [TT]-av[tt]" },
{ "-integrate", FALSE, etBOOL, {&bIntegrate},
"Integrate data function(s) numerically using trapezium rule" },
{ "-aver_start", FALSE, etREAL, {&aver_start},
"Start averaging the integral from here" },
{ "-xydy", FALSE, etBOOL, {&bXYdy},
"Interpret second data set as error in the y values for integrating" },
{ "-regression", FALSE, etBOOL, {&bRegression},
"Perform a linear regression analysis on the data. If [TT]-xydy[tt] is set a second set will be interpreted as the error bar in the Y value. Otherwise, if multiple data sets are present a multilinear regression will be performed yielding the constant A that minimize [MATH][GRK]chi[grk]^2 = (y - A[SUB]0[sub] x[SUB]0[sub] - A[SUB]1[sub] x[SUB]1[sub] - ... - A[SUB]N[sub] x[SUB]N[sub])^2[math] where now Y is the first data set in the input file and x[SUB]i[sub] the others. Do read the information at the option [TT]-time[tt]." },
{ "-luzar", FALSE, etBOOL, {&bLuzar},
"Do a Luzar and Chandler analysis on a correlation function and related as produced by [TT]g_hbond[tt]. When in addition the [TT]-xydy[tt] flag is given the second and fourth column will be interpreted as errors in c(t) and n(t)." },
{ "-temp", FALSE, etREAL, {&temp},
"Temperature for the Luzar hydrogen bonding kinetics analysis (K)" },
{ "-fitstart", FALSE, etREAL, {&fit_start},
"Time (ps) from which to start fitting the correlation functions in order to obtain the forward and backward rate constants for HB breaking and formation" },
{ "-fitend", FALSE, etREAL, {&fit_end},
"Time (ps) where to stop fitting the correlation functions in order to obtain the forward and backward rate constants for HB breaking and formation. Only with [TT]-gem[tt]" },
{ "-smooth", FALSE, etREAL, {&smooth_tail_start},
"If this value is >= 0, the tail of the ACF will be smoothed by fitting it to an exponential function: [MATH]y = A [EXP]-x/[GRK]tau[grk][exp][math]" },
{ "-nbmin", FALSE, etINT, {&nb_min},
"HIDDENMinimum number of blocks for block averaging" },
{ "-resol", FALSE, etINT, {&resol},
"HIDDENResolution for the block averaging, block size increases with"
" a factor 2^(1/resol)" },
{ "-eeexpfit", FALSE, etBOOL, {&bEESEF},
"HIDDENAlways use a single exponential fit for the error estimate" },
{ "-eenlc", FALSE, etBOOL, {&bEENLC},
"HIDDENAllow a negative long-time correlation" },
{ "-eefitac", FALSE, etBOOL, {&bEeFitAc},
"HIDDENAlso plot analytical block average using a autocorrelation fit" },
{ "-filter", FALSE, etREAL, {&filtlen},
"Print the high-frequency fluctuation after filtering with a cosine filter of this length" },
{ "-power", FALSE, etBOOL, {&bPower},
"Fit data to: b t^a" },
{ "-subav", FALSE, etBOOL, {&bSubAv},
"Subtract the average before autocorrelating" },
{ "-oneacf", FALSE, etBOOL, {&bAverCorr},
"Calculate one ACF over all sets" },
{ "-nbalexp", FALSE, etINT, {&nBalExp},
"HIDDENNumber of exponentials to fit to the ultrafast component" },
{ "-baltime", FALSE, etREAL, {&balTime},
"HIDDENTime up to which the ballistic component will be fitted" },
/* { "-gemnp", FALSE, etINT, {&nFitPoints}, */
/* "HIDDENNumber of data points taken from the ACF to use for fitting to rev. gem. recomb. model."}, */
/* { "-rcut", FALSE, etREAL, {&rcut}, */
/* "Cut-off for hydrogen bonds in geminate algorithms" }, */
/* { "-gemtype", FALSE, etENUM, {gemType}, */
/* "What type of gminate recombination to use"}, */
/* { "-D", FALSE, etREAL, {&diffusion}, */
/* "The self diffusion coefficient which is used for the reversible geminate recombination model."} */
};
#define NPA asize(pa)
FILE *out, *out_fit;
int n, nlast, s, nset, i, j = 0;
real **val, *t, dt, tot, error;
double *av, *sig, cum1, cum2, cum3, cum4, db;
const char *acfile, *msdfile, *ccfile, *distfile, *avfile, *eefile, *balfile, *gemfile, *fitfile;
output_env_t oenv;
t_filenm fnm[] = {
{ efXVG, "-f", "graph", ffREAD },
{ efXVG, "-ac", "autocorr", ffOPTWR },
{ efXVG, "-msd", "msd", ffOPTWR },
{ efXVG, "-cc", "coscont", ffOPTWR },
{ efXVG, "-dist", "distr", ffOPTWR },
{ efXVG, "-av", "average", ffOPTWR },
{ efXVG, "-ee", "errest", ffOPTWR },
{ efXVG, "-bal", "ballisitc", ffOPTWR },
/* { efXVG, "-gem", "geminate", ffOPTWR }, */
{ efLOG, "-g", "fitlog", ffOPTWR }
};
#define NFILE asize(fnm)
int npargs;
t_pargs *ppa;
npargs = asize(pa);
ppa = add_acf_pargs(&npargs, pa);
parse_common_args(&argc, argv, PCA_CAN_VIEW,
NFILE, fnm, npargs, ppa, asize(desc), desc, 0, NULL, &oenv);
acfile = opt2fn_null("-ac", NFILE, fnm);
msdfile = opt2fn_null("-msd", NFILE, fnm);
ccfile = opt2fn_null("-cc", NFILE, fnm);
distfile = opt2fn_null("-dist", NFILE, fnm);
avfile = opt2fn_null("-av", NFILE, fnm);
eefile = opt2fn_null("-ee", NFILE, fnm);
balfile = opt2fn_null("-bal", NFILE, fnm);
/* gemfile = opt2fn_null("-gem",NFILE,fnm); */
/* When doing autocorrelation we don't want a fitlog for fitting
* the function itself (not the acf) when the user did not ask for it.
*/
if (opt2parg_bSet("-fitfn", npargs, ppa) && acfile == NULL)
{
fitfile = opt2fn("-g", NFILE, fnm);
}
else
{
fitfile = opt2fn_null("-g", NFILE, fnm);
}
val = read_xvg_time(opt2fn("-f", NFILE, fnm), bHaveT,
opt2parg_bSet("-b", npargs, ppa), tb,
opt2parg_bSet("-e", npargs, ppa), te,
nsets_in, &nset, &n, &dt, &t);
printf("Read %d sets of %d points, dt = %g\n\n", nset, n, dt);
if (bDer)
{
printf("Calculating the derivative as (f[i+%d]-f[i])/(%d*dt)\n\n",
d, d);
n -= d;
for (s = 0; s < nset; s++)
{
for (i = 0; (i < n); i++)
{
val[s][i] = (val[s][i+d]-val[s][i])/(d*dt);
}
}
}
if (bIntegrate)
{
real sum, stddev;
printf("Calculating the integral using the trapezium rule\n");
if (bXYdy)
{
sum = evaluate_integral(n, t, val[0], val[1], aver_start, &stddev);
printf("Integral %10.3f +/- %10.5f\n", sum, stddev);
}
else
{
for (s = 0; s < nset; s++)
{
sum = evaluate_integral(n, t, val[s], NULL, aver_start, &stddev);
printf("Integral %d %10.5f +/- %10.5f\n", s+1, sum, stddev);
}
}
}
if (fitfile != NULL)
{
out_fit = ffopen(fitfile, "w");
if (bXYdy && nset >= 2)
{
do_fit(out_fit, 0, TRUE, n, t, val, npargs, ppa, oenv);
}
else
{
for (s = 0; s < nset; s++)
{
do_fit(out_fit, s, FALSE, n, t, val, npargs, ppa, oenv);
}
}
ffclose(out_fit);
}
printf(" std. dev. relative deviation of\n");
printf(" standard --------- cumulants from those of\n");
printf("set average deviation sqrt(n-1) a Gaussian distribition\n");
printf(" cum. 3 cum. 4\n");
snew(av, nset);
snew(sig, nset);
for (s = 0; (s < nset); s++)
{
cum1 = 0;
cum2 = 0;
cum3 = 0;
cum4 = 0;
for (i = 0; (i < n); i++)
{
cum1 += val[s][i];
}
cum1 /= n;
for (i = 0; (i < n); i++)
{
db = val[s][i]-cum1;
cum2 += db*db;
cum3 += db*db*db;
cum4 += db*db*db*db;
}
cum2 /= n;
cum3 /= n;
cum4 /= n;
av[s] = cum1;
sig[s] = sqrt(cum2);
if (n > 1)
{
error = sqrt(cum2/(n-1));
}
else
{
error = 0;
}
printf("SS%d %13.6e %12.6e %12.6e %6.3f %6.3f\n",
s+1, av[s], sig[s], error,
sig[s] ? cum3/(sig[s]*sig[s]*sig[s]*sqrt(8/M_PI)) : 0,
sig[s] ? cum4/(sig[s]*sig[s]*sig[s]*sig[s]*3)-1 : 0);
}
printf("\n");
if (filtlen)
{
filter(filtlen, n, nset, val, dt);
}
if (msdfile)
{
out = xvgropen(msdfile, "Mean square displacement",
"time", "MSD (nm\\S2\\N)", oenv);
nlast = (int)(n*frac);
for (s = 0; s < nset; s++)
{
for (j = 0; j <= nlast; j++)
{
if (j % 100 == 0)
{
fprintf(stderr, "\r%d", j);
}
tot = 0;
for (i = 0; i < n-j; i++)
{
tot += sqr(val[s][i]-val[s][i+j]);
}
tot /= (real)(n-j);
fprintf(out, " %g %8g\n", dt*j, tot);
}
if (s < nset-1)
{
fprintf(out, "&\n");
}
}
ffclose(out);
fprintf(stderr, "\r%d, time=%g\n", j-1, (j-1)*dt);
}
if (ccfile)
{
plot_coscont(ccfile, n, nset, val, oenv);
}
if (distfile)
{
histogram(distfile, binwidth, n, nset, val, oenv);
}
if (avfile)
{
average(avfile, nenum(avbar_opt), n, nset, val, t);
}
if (eefile)
{
estimate_error(eefile, nb_min, resol, n, nset, av, sig, val, dt,
bEeFitAc, bEESEF, bEENLC, oenv);
}
if (balfile)
{
do_ballistic(balfile, n, t, val, nset, balTime, nBalExp, oenv);
}
/* if (gemfile) */
/* do_geminate(gemfile,n,t,val,nset,diffusion,rcut,balTime, */
/* nFitPoints, fit_start, fit_end, oenv); */
if (bPower)
{
power_fit(n, nset, val, t);
}
if (acfile != NULL)
{
if (bSubAv)
{
for (s = 0; s < nset; s++)
{
for (i = 0; i < n; i++)
{
val[s][i] -= av[s];
}
}
}
do_autocorr(acfile, oenv, "Autocorrelation", n, nset, val, dt,
eacNormal, bAverCorr);
}
if (bRegression)
{
regression_analysis(n, bXYdy, t, nset, val);
}
if (bLuzar)
{
luzar_correl(n, t, nset, val, temp, bXYdy, fit_start, smooth_tail_start, oenv);
}
view_all(oenv, NFILE, fnm);
return 0;
}
int gmx_rms(int argc, char *argv[])
{
const char *desc[] =
{
"[THISMODULE] compares two structures by computing the root mean square",
"deviation (RMSD), the size-independent [GRK]rho[grk] similarity parameter",
"([TT]rho[tt]) or the scaled [GRK]rho[grk] ([TT]rhosc[tt]), ",
"see Maiorov & Crippen, Proteins [BB]22[bb], 273 (1995).",
"This is selected by [TT]-what[tt].[PAR]"
"Each structure from a trajectory ([TT]-f[tt]) is compared to a",
"reference structure. The reference structure",
"is taken from the structure file ([TT]-s[tt]).[PAR]",
"With option [TT]-mir[tt] also a comparison with the mirror image of",
"the reference structure is calculated.",
"This is useful as a reference for 'significant' values, see",
"Maiorov & Crippen, Proteins [BB]22[bb], 273 (1995).[PAR]",
"Option [TT]-prev[tt] produces the comparison with a previous frame",
"the specified number of frames ago.[PAR]",
"Option [TT]-m[tt] produces a matrix in [TT].xpm[tt] format of",
"comparison values of each structure in the trajectory with respect to",
"each other structure. This file can be visualized with for instance",
"[TT]xv[tt] and can be converted to postscript with [gmx-xpm2ps].[PAR]",
"Option [TT]-fit[tt] controls the least-squares fitting of",
"the structures on top of each other: complete fit (rotation and",
"translation), translation only, or no fitting at all.[PAR]",
"Option [TT]-mw[tt] controls whether mass weighting is done or not.",
"If you select the option (default) and ",
"supply a valid [TT].tpr[tt] file masses will be taken from there, ",
"otherwise the masses will be deduced from the [TT]atommass.dat[tt] file in",
"[TT]GMXLIB[tt]. This is fine for proteins, but not",
"necessarily for other molecules. A default mass of 12.011 amu (carbon)",
"is assigned to unknown atoms. You can check whether this happend by",
"turning on the [TT]-debug[tt] flag and inspecting the log file.[PAR]",
"With [TT]-f2[tt], the 'other structures' are taken from a second",
"trajectory, this generates a comparison matrix of one trajectory",
"versus the other.[PAR]",
"Option [TT]-bin[tt] does a binary dump of the comparison matrix.[PAR]",
"Option [TT]-bm[tt] produces a matrix of average bond angle deviations",
"analogously to the [TT]-m[tt] option. Only bonds between atoms in the",
"comparison group are considered."
};
static gmx_bool bPBC = TRUE, bFitAll = TRUE, bSplit = FALSE;
static gmx_bool bDeltaLog = FALSE;
static int prev = 0, freq = 1, freq2 = 1, nlevels = 80, avl = 0;
static real rmsd_user_max = -1, rmsd_user_min = -1, bond_user_max = -1,
bond_user_min = -1, delta_maxy = 0.0;
/* strings and things for selecting difference method */
enum
{
ewSel, ewRMSD, ewRho, ewRhoSc, ewNR
};
int ewhat;
const char *what[ewNR + 1] =
{ NULL, "rmsd", "rho", "rhosc", NULL };
const char *whatname[ewNR] =
{ NULL, "RMSD", "Rho", "Rho sc" };
const char *whatlabel[ewNR] =
{ NULL, "RMSD (nm)", "Rho", "Rho sc" };
const char *whatxvgname[ewNR] =
{ NULL, "RMSD", "\\8r\\4", "\\8r\\4\\ssc\\N" };
const char *whatxvglabel[ewNR] =
{ NULL, "RMSD (nm)", "\\8r\\4", "\\8r\\4\\ssc\\N" };
/* strings and things for fitting methods */
enum
{
efSel, efFit, efReset, efNone, efNR
};
int efit;
const char *fit[efNR + 1] =
{ NULL, "rot+trans", "translation", "none", NULL };
const char *fitgraphlabel[efNR + 1] =
{ NULL, "lsq fit", "translational fit", "no fit" };
static int nrms = 1;
static gmx_bool bMassWeighted = TRUE;
t_pargs pa[] =
{
{ "-what", FALSE, etENUM,
{ what }, "Structural difference measure" },
{ "-pbc", FALSE, etBOOL,
{ &bPBC }, "PBC check" },
{ "-fit", FALSE, etENUM,
{ fit }, "Fit to reference structure" },
{ "-prev", FALSE, etINT,
{ &prev }, "Compare with previous frame" },
{ "-split", FALSE, etBOOL,
{ &bSplit }, "Split graph where time is zero" },
{ "-fitall", FALSE, etBOOL,
{ &bFitAll }, "HIDDENFit all pairs of structures in matrix" },
{ "-skip", FALSE, etINT,
{ &freq }, "Only write every nr-th frame to matrix" },
{ "-skip2", FALSE, etINT,
{ &freq2 }, "Only write every nr-th frame to matrix" },
{ "-max", FALSE, etREAL,
{ &rmsd_user_max }, "Maximum level in comparison matrix" },
{ "-min", FALSE, etREAL,
{ &rmsd_user_min }, "Minimum level in comparison matrix" },
{ "-bmax", FALSE, etREAL,
{ &bond_user_max }, "Maximum level in bond angle matrix" },
{ "-bmin", FALSE, etREAL,
{ &bond_user_min }, "Minimum level in bond angle matrix" },
{ "-mw", FALSE, etBOOL,
{ &bMassWeighted }, "Use mass weighting for superposition" },
{ "-nlevels", FALSE, etINT,
{ &nlevels }, "Number of levels in the matrices" },
{ "-ng", FALSE, etINT,
{ &nrms }, "Number of groups to compute RMS between" },
{ "-dlog", FALSE, etBOOL,
{ &bDeltaLog },
"HIDDENUse a log x-axis in the delta t matrix" },
{ "-dmax", FALSE, etREAL,
{ &delta_maxy }, "HIDDENMaximum level in delta matrix" },
{ "-aver", FALSE, etINT,
{ &avl },
"HIDDENAverage over this distance in the RMSD matrix" }
};
int natoms_trx, natoms_trx2, natoms;
int i, j, k, m, teller, teller2, tel_mat, tel_mat2;
#define NFRAME 5000
int maxframe = NFRAME, maxframe2 = NFRAME;
real t, *w_rls, *w_rms, *w_rls_m = NULL, *w_rms_m = NULL;
gmx_bool bNorm, bAv, bFreq2, bFile2, bMat, bBond, bDelta, bMirror, bMass;
gmx_bool bFit, bReset;
t_topology top;
int ePBC;
t_iatom *iatom = NULL;
matrix box;
rvec *x, *xp, *xm = NULL, **mat_x = NULL, **mat_x2, *mat_x2_j = NULL, vec1,
vec2;
t_trxstatus *status;
char buf[256], buf2[256];
int ncons = 0;
FILE *fp;
real rlstot = 0, **rls, **rlsm = NULL, *time, *time2, *rlsnorm = NULL,
**rmsd_mat = NULL, **bond_mat = NULL, *axis, *axis2, *del_xaxis,
*del_yaxis, rmsd_max, rmsd_min, rmsd_avg, bond_max, bond_min, ang;
real **rmsdav_mat = NULL, av_tot, weight, weight_tot;
real **delta = NULL, delta_max, delta_scalex = 0, delta_scaley = 0,
*delta_tot;
int delta_xsize = 0, del_lev = 100, mx, my, abs_my;
gmx_bool bA1, bA2, bPrev, bTop, *bInMat = NULL;
int ifit, *irms, ibond = 0, *ind_bond1 = NULL, *ind_bond2 = NULL, n_ind_m =
0;
atom_id *ind_fit, **ind_rms, *ind_m = NULL, *rev_ind_m = NULL, *ind_rms_m =
NULL;
char *gn_fit, **gn_rms;
t_rgb rlo, rhi;
output_env_t oenv;
gmx_rmpbc_t gpbc = NULL;
t_filenm fnm[] =
{
{ efTPS, NULL, NULL, ffREAD },
{ efTRX, "-f", NULL, ffREAD },
{ efTRX, "-f2", NULL, ffOPTRD },
{ efNDX, NULL, NULL, ffOPTRD },
{ efXVG, NULL, "rmsd", ffWRITE },
{ efXVG, "-mir", "rmsdmir", ffOPTWR },
{ efXVG, "-a", "avgrp", ffOPTWR },
{ efXVG, "-dist", "rmsd-dist", ffOPTWR },
{ efXPM, "-m", "rmsd", ffOPTWR },
{ efDAT, "-bin", "rmsd", ffOPTWR },
{ efXPM, "-bm", "bond", ffOPTWR }
};
#define NFILE asize(fnm)
if (!parse_common_args(&argc, argv, PCA_CAN_TIME | PCA_TIME_UNIT | PCA_CAN_VIEW
| PCA_BE_NICE, NFILE, fnm, asize(pa), pa, asize(desc), desc, 0, NULL,
&oenv))
{
return 0;
}
/* parse enumerated options: */
ewhat = nenum(what);
if (ewhat == ewRho || ewhat == ewRhoSc)
{
please_cite(stdout, "Maiorov95");
}
efit = nenum(fit);
bFit = efit == efFit;
bReset = efit == efReset;
if (bFit)
{
bReset = TRUE; /* for fit, reset *must* be set */
}
else
{
bFitAll = FALSE;
}
/* mark active cmdline options */
bMirror = opt2bSet("-mir", NFILE, fnm); /* calc RMSD vs mirror of ref. */
bFile2 = opt2bSet("-f2", NFILE, fnm);
bMat = opt2bSet("-m", NFILE, fnm);
bBond = opt2bSet("-bm", NFILE, fnm);
bDelta = (delta_maxy > 0); /* calculate rmsd vs delta t matrix from *
* your RMSD matrix (hidden option */
bNorm = opt2bSet("-a", NFILE, fnm);
bFreq2 = opt2parg_bSet("-skip2", asize(pa), pa);
if (freq <= 0)
{
fprintf(stderr, "The number of frames to skip is <= 0. "
"Writing out all frames.\n\n");
freq = 1;
}
if (!bFreq2)
{
freq2 = freq;
}
else if (bFile2 && freq2 <= 0)
{
fprintf(stderr,
"The number of frames to skip in second trajectory is <= 0.\n"
" Writing out all frames.\n\n");
freq2 = 1;
}
bPrev = (prev > 0);
if (bPrev)
{
prev = abs(prev);
if (freq != 1)
{
fprintf(stderr, "WARNING: option -skip also applies to -prev\n");
}
}
if (bFile2 && !bMat && !bBond)
{
fprintf(
stderr,
"WARNING: second trajectory (-f2) useless when not calculating matrix (-m/-bm),\n"
" will not read from %s\n", opt2fn("-f2", NFILE,
fnm));
bFile2 = FALSE;
}
if (bDelta)
{
bMat = TRUE;
if (bFile2)
{
fprintf(stderr,
"WARNING: second trajectory (-f2) useless when making delta matrix,\n"
" will not read from %s\n", opt2fn("-f2",
NFILE, fnm));
bFile2 = FALSE;
}
}
bTop = read_tps_conf(ftp2fn(efTPS, NFILE, fnm), buf, &top, &ePBC, &xp,
NULL, box, TRUE);
snew(w_rls, top.atoms.nr);
snew(w_rms, top.atoms.nr);
if (!bTop && bBond)
{
fprintf(stderr,
"WARNING: Need a run input file for bond angle matrix,\n"
" will not calculate bond angle matrix.\n");
bBond = FALSE;
}
if (bReset)
{
fprintf(stderr, "Select group for %s fit\n", bFit ? "least squares"
: "translational");
get_index(&(top.atoms), ftp2fn_null(efNDX, NFILE, fnm), 1, &ifit,
&ind_fit, &gn_fit);
}
else
{
ifit = 0;
}
if (bReset)
{
if (bFit && ifit < 3)
{
gmx_fatal(FARGS, "Need >= 3 points to fit!\n" );
}
bMass = FALSE;
for (i = 0; i < ifit; i++)
{
if (bMassWeighted)
{
w_rls[ind_fit[i]] = top.atoms.atom[ind_fit[i]].m;
}
else
{
w_rls[ind_fit[i]] = 1;
}
bMass = bMass || (top.atoms.atom[ind_fit[i]].m != 0);
}
if (!bMass)
{
fprintf(stderr, "All masses in the fit group are 0, using masses of 1\n");
for (i = 0; i < ifit; i++)
{
w_rls[ind_fit[i]] = 1;
}
}
}
if (bMat || bBond)
{
nrms = 1;
}
snew(gn_rms, nrms);
snew(ind_rms, nrms);
snew(irms, nrms);
fprintf(stderr, "Select group%s for %s calculation\n",
(nrms > 1) ? "s" : "", whatname[ewhat]);
get_index(&(top.atoms), ftp2fn_null(efNDX, NFILE, fnm),
nrms, irms, ind_rms, gn_rms);
if (bNorm)
{
snew(rlsnorm, irms[0]);
}
snew(rls, nrms);
for (j = 0; j < nrms; j++)
{
snew(rls[j], maxframe);
}
if (bMirror)
{
snew(rlsm, nrms);
for (j = 0; j < nrms; j++)
{
snew(rlsm[j], maxframe);
}
}
snew(time, maxframe);
for (j = 0; j < nrms; j++)
{
bMass = FALSE;
for (i = 0; i < irms[j]; i++)
{
if (bMassWeighted)
{
w_rms[ind_rms[j][i]] = top.atoms.atom[ind_rms[j][i]].m;
}
else
{
w_rms[ind_rms[j][i]] = 1.0;
}
bMass = bMass || (top.atoms.atom[ind_rms[j][i]].m != 0);
}
if (!bMass)
{
fprintf(stderr, "All masses in group %d are 0, using masses of 1\n", j);
for (i = 0; i < irms[j]; i++)
{
w_rms[ind_rms[j][i]] = 1;
}
}
}
/* Prepare reference frame */
if (bPBC)
{
gpbc = gmx_rmpbc_init(&top.idef, ePBC, top.atoms.nr);
gmx_rmpbc(gpbc, top.atoms.nr, box, xp);
}
if (bReset)
{
reset_x(ifit, ind_fit, top.atoms.nr, NULL, xp, w_rls);
}
if (bMirror)
{
/* generate reference structure mirror image: */
snew(xm, top.atoms.nr);
for (i = 0; i < top.atoms.nr; i++)
{
copy_rvec(xp[i], xm[i]);
xm[i][XX] = -xm[i][XX];
}
}
if (ewhat == ewRhoSc)
{
norm_princ(&top.atoms, ifit, ind_fit, top.atoms.nr, xp);
}
/* read first frame */
natoms_trx = read_first_x(oenv, &status, opt2fn("-f", NFILE, fnm), &t, &x, box);
if (natoms_trx != top.atoms.nr)
{
fprintf(stderr,
"\nWARNING: topology has %d atoms, whereas trajectory has %d\n",
top.atoms.nr, natoms_trx);
}
natoms = min(top.atoms.nr, natoms_trx);
if (bMat || bBond || bPrev)
{
snew(mat_x, NFRAME);
if (bPrev)
{
/* With -prev we use all atoms for simplicity */
n_ind_m = natoms;
}
else
{
/* Check which atoms we need (fit/rms) */
snew(bInMat, natoms);
for (i = 0; i < ifit; i++)
{
bInMat[ind_fit[i]] = TRUE;
}
n_ind_m = ifit;
for (i = 0; i < irms[0]; i++)
{
if (!bInMat[ind_rms[0][i]])
{
bInMat[ind_rms[0][i]] = TRUE;
n_ind_m++;
}
}
}
/* Make an index of needed atoms */
snew(ind_m, n_ind_m);
snew(rev_ind_m, natoms);
j = 0;
for (i = 0; i < natoms; i++)
{
if (bPrev || bInMat[i])
{
ind_m[j] = i;
rev_ind_m[i] = j;
j++;
}
}
snew(w_rls_m, n_ind_m);
snew(ind_rms_m, irms[0]);
snew(w_rms_m, n_ind_m);
for (i = 0; i < ifit; i++)
{
w_rls_m[rev_ind_m[ind_fit[i]]] = w_rls[ind_fit[i]];
}
for (i = 0; i < irms[0]; i++)
{
ind_rms_m[i] = rev_ind_m[ind_rms[0][i]];
w_rms_m[ind_rms_m[i]] = w_rms[ind_rms[0][i]];
}
sfree(bInMat);
}
if (bBond)
{
ncons = 0;
for (k = 0; k < F_NRE; k++)
{
if (IS_CHEMBOND(k))
{
iatom = top.idef.il[k].iatoms;
ncons += top.idef.il[k].nr/3;
}
}
fprintf(stderr, "Found %d bonds in topology\n", ncons);
snew(ind_bond1, ncons);
snew(ind_bond2, ncons);
ibond = 0;
for (k = 0; k < F_NRE; k++)
{
if (IS_CHEMBOND(k))
{
iatom = top.idef.il[k].iatoms;
ncons = top.idef.il[k].nr/3;
for (i = 0; i < ncons; i++)
{
bA1 = FALSE;
bA2 = FALSE;
for (j = 0; j < irms[0]; j++)
{
if (iatom[3*i+1] == ind_rms[0][j])
{
bA1 = TRUE;
}
if (iatom[3*i+2] == ind_rms[0][j])
{
bA2 = TRUE;
}
}
if (bA1 && bA2)
{
ind_bond1[ibond] = rev_ind_m[iatom[3*i+1]];
ind_bond2[ibond] = rev_ind_m[iatom[3*i+2]];
ibond++;
}
}
}
}
fprintf(stderr, "Using %d bonds for bond angle matrix\n", ibond);
if (ibond == 0)
{
gmx_fatal(FARGS, "0 bonds found");
}
}
/* start looping over frames: */
tel_mat = 0;
teller = 0;
do
{
if (bPBC)
{
gmx_rmpbc(gpbc, natoms, box, x);
}
if (bReset)
{
reset_x(ifit, ind_fit, natoms, NULL, x, w_rls);
}
if (ewhat == ewRhoSc)
{
norm_princ(&top.atoms, ifit, ind_fit, natoms, x);
}
if (bFit)
{
/*do the least squares fit to original structure*/
do_fit(natoms, w_rls, xp, x);
}
if (teller % freq == 0)
{
/* keep frame for matrix calculation */
if (bMat || bBond || bPrev)
{
if (tel_mat >= NFRAME)
{
srenew(mat_x, tel_mat+1);
}
snew(mat_x[tel_mat], n_ind_m);
for (i = 0; i < n_ind_m; i++)
{
copy_rvec(x[ind_m[i]], mat_x[tel_mat][i]);
}
}
tel_mat++;
}
/*calculate energy of root_least_squares*/
if (bPrev)
{
j = tel_mat-prev-1;
if (j < 0)
{
j = 0;
}
for (i = 0; i < n_ind_m; i++)
{
copy_rvec(mat_x[j][i], xp[ind_m[i]]);
}
if (bReset)
{
reset_x(ifit, ind_fit, natoms, NULL, xp, w_rls);
}
if (bFit)
{
do_fit(natoms, w_rls, x, xp);
}
}
for (j = 0; (j < nrms); j++)
{
rls[j][teller] =
calc_similar_ind(ewhat != ewRMSD, irms[j], ind_rms[j], w_rms, x, xp);
}
if (bNorm)
{
for (j = 0; (j < irms[0]); j++)
{
rlsnorm[j] +=
calc_similar_ind(ewhat != ewRMSD, 1, &(ind_rms[0][j]), w_rms, x, xp);
}
}
if (bMirror)
{
if (bFit)
{
/*do the least squares fit to mirror of original structure*/
do_fit(natoms, w_rls, xm, x);
}
for (j = 0; j < nrms; j++)
{
rlsm[j][teller] =
calc_similar_ind(ewhat != ewRMSD, irms[j], ind_rms[j], w_rms, x, xm);
}
}
time[teller] = output_env_conv_time(oenv, t);
teller++;
if (teller >= maxframe)
{
maxframe += NFRAME;
srenew(time, maxframe);
for (j = 0; (j < nrms); j++)
{
srenew(rls[j], maxframe);
}
if (bMirror)
{
for (j = 0; (j < nrms); j++)
{
srenew(rlsm[j], maxframe);
}
}
}
}
while (read_next_x(oenv, status, &t, x, box));
close_trj(status);
if (bFile2)
{
snew(time2, maxframe2);
fprintf(stderr, "\nWill read second trajectory file\n");
snew(mat_x2, NFRAME);
natoms_trx2 =
read_first_x(oenv, &status, opt2fn("-f2", NFILE, fnm), &t, &x, box);
if (natoms_trx2 != natoms_trx)
{
gmx_fatal(FARGS,
"Second trajectory (%d atoms) does not match the first one"
" (%d atoms)", natoms_trx2, natoms_trx);
}
tel_mat2 = 0;
teller2 = 0;
do
{
if (bPBC)
{
gmx_rmpbc(gpbc, natoms, box, x);
}
if (bReset)
{
reset_x(ifit, ind_fit, natoms, NULL, x, w_rls);
}
if (ewhat == ewRhoSc)
{
norm_princ(&top.atoms, ifit, ind_fit, natoms, x);
}
if (bFit)
{
/*do the least squares fit to original structure*/
do_fit(natoms, w_rls, xp, x);
}
if (teller2 % freq2 == 0)
{
/* keep frame for matrix calculation */
if (bMat)
{
if (tel_mat2 >= NFRAME)
{
srenew(mat_x2, tel_mat2+1);
}
snew(mat_x2[tel_mat2], n_ind_m);
for (i = 0; i < n_ind_m; i++)
{
copy_rvec(x[ind_m[i]], mat_x2[tel_mat2][i]);
}
}
tel_mat2++;
}
time2[teller2] = output_env_conv_time(oenv, t);
teller2++;
if (teller2 >= maxframe2)
{
maxframe2 += NFRAME;
srenew(time2, maxframe2);
}
}
while (read_next_x(oenv, status, &t, x, box));
close_trj(status);
}
else
{
mat_x2 = mat_x;
time2 = time;
tel_mat2 = tel_mat;
freq2 = freq;
}
gmx_rmpbc_done(gpbc);
if (bMat || bBond)
{
/* calculate RMS matrix */
fprintf(stderr, "\n");
if (bMat)
{
fprintf(stderr, "Building %s matrix, %dx%d elements\n",
whatname[ewhat], tel_mat, tel_mat2);
snew(rmsd_mat, tel_mat);
}
if (bBond)
{
fprintf(stderr, "Building bond angle matrix, %dx%d elements\n",
tel_mat, tel_mat2);
snew(bond_mat, tel_mat);
}
snew(axis, tel_mat);
snew(axis2, tel_mat2);
rmsd_max = 0;
if (bFile2)
{
rmsd_min = 1e10;
}
else
{
rmsd_min = 0;
}
rmsd_avg = 0;
bond_max = 0;
bond_min = 1e10;
for (j = 0; j < tel_mat2; j++)
{
axis2[j] = time2[freq2*j];
}
if (bDelta)
{
if (bDeltaLog)
{
delta_scalex = 8.0/log(2.0);
delta_xsize = (int)(log(tel_mat/2)*delta_scalex+0.5)+1;
}
else
{
delta_xsize = tel_mat/2;
}
delta_scaley = 1.0/delta_maxy;
snew(delta, delta_xsize);
for (j = 0; j < delta_xsize; j++)
{
snew(delta[j], del_lev+1);
}
if (avl > 0)
{
snew(rmsdav_mat, tel_mat);
for (j = 0; j < tel_mat; j++)
{
snew(rmsdav_mat[j], tel_mat);
}
}
}
if (bFitAll)
{
snew(mat_x2_j, natoms);
}
for (i = 0; i < tel_mat; i++)
{
axis[i] = time[freq*i];
fprintf(stderr, "\r element %5d; time %5.2f ", i, axis[i]);
if (bMat)
{
snew(rmsd_mat[i], tel_mat2);
}
if (bBond)
{
snew(bond_mat[i], tel_mat2);
}
for (j = 0; j < tel_mat2; j++)
{
if (bFitAll)
{
for (k = 0; k < n_ind_m; k++)
{
copy_rvec(mat_x2[j][k], mat_x2_j[k]);
}
do_fit(n_ind_m, w_rls_m, mat_x[i], mat_x2_j);
}
else
{
mat_x2_j = mat_x2[j];
}
if (bMat)
{
if (bFile2 || (i < j))
{
rmsd_mat[i][j] =
calc_similar_ind(ewhat != ewRMSD, irms[0], ind_rms_m,
w_rms_m, mat_x[i], mat_x2_j);
if (rmsd_mat[i][j] > rmsd_max)
{
rmsd_max = rmsd_mat[i][j];
}
if (rmsd_mat[i][j] < rmsd_min)
{
rmsd_min = rmsd_mat[i][j];
}
rmsd_avg += rmsd_mat[i][j];
}
else
{
rmsd_mat[i][j] = rmsd_mat[j][i];
}
}
if (bBond)
{
if (bFile2 || (i <= j))
{
ang = 0.0;
for (m = 0; m < ibond; m++)
{
rvec_sub(mat_x[i][ind_bond1[m]], mat_x[i][ind_bond2[m]], vec1);
rvec_sub(mat_x2_j[ind_bond1[m]], mat_x2_j[ind_bond2[m]], vec2);
ang += acos(cos_angle(vec1, vec2));
}
bond_mat[i][j] = ang*180.0/(M_PI*ibond);
if (bond_mat[i][j] > bond_max)
{
bond_max = bond_mat[i][j];
}
if (bond_mat[i][j] < bond_min)
{
bond_min = bond_mat[i][j];
}
}
else
{
bond_mat[i][j] = bond_mat[j][i];
}
}
}
}
if (bFile2)
{
rmsd_avg /= tel_mat*tel_mat2;
}
else
{
rmsd_avg /= tel_mat*(tel_mat - 1)/2;
}
if (bMat && (avl > 0))
{
rmsd_max = 0.0;
rmsd_min = 0.0;
rmsd_avg = 0.0;
for (j = 0; j < tel_mat-1; j++)
{
for (i = j+1; i < tel_mat; i++)
{
av_tot = 0;
weight_tot = 0;
for (my = -avl; my <= avl; my++)
{
if ((j+my >= 0) && (j+my < tel_mat))
{
abs_my = abs(my);
for (mx = -avl; mx <= avl; mx++)
{
if ((i+mx >= 0) && (i+mx < tel_mat))
{
weight = (real)(avl+1-max(abs(mx), abs_my));
av_tot += weight*rmsd_mat[i+mx][j+my];
weight_tot += weight;
}
}
}
}
rmsdav_mat[i][j] = av_tot/weight_tot;
rmsdav_mat[j][i] = rmsdav_mat[i][j];
if (rmsdav_mat[i][j] > rmsd_max)
{
rmsd_max = rmsdav_mat[i][j];
}
}
}
rmsd_mat = rmsdav_mat;
}
if (bMat)
{
fprintf(stderr, "\n%s: Min %f, Max %f, Avg %f\n",
whatname[ewhat], rmsd_min, rmsd_max, rmsd_avg);
rlo.r = 1; rlo.g = 1; rlo.b = 1;
rhi.r = 0; rhi.g = 0; rhi.b = 0;
if (rmsd_user_max != -1)
{
rmsd_max = rmsd_user_max;
}
if (rmsd_user_min != -1)
{
rmsd_min = rmsd_user_min;
}
if ((rmsd_user_max != -1) || (rmsd_user_min != -1))
{
fprintf(stderr, "Min and Max value set to resp. %f and %f\n",
rmsd_min, rmsd_max);
}
sprintf(buf, "%s %s matrix", gn_rms[0], whatname[ewhat]);
write_xpm(opt2FILE("-m", NFILE, fnm, "w"), 0, buf, whatlabel[ewhat],
output_env_get_time_label(oenv), output_env_get_time_label(oenv), tel_mat, tel_mat2,
axis, axis2, rmsd_mat, rmsd_min, rmsd_max, rlo, rhi, &nlevels);
/* Print the distribution of RMSD values */
if (opt2bSet("-dist", NFILE, fnm))
{
low_rmsd_dist(opt2fn("-dist", NFILE, fnm), rmsd_max, tel_mat, rmsd_mat, oenv);
}
if (bDelta)
{
snew(delta_tot, delta_xsize);
for (j = 0; j < tel_mat-1; j++)
{
for (i = j+1; i < tel_mat; i++)
{
mx = i-j;
if (mx < tel_mat/2)
{
if (bDeltaLog)
{
mx = (int)(log(mx)*delta_scalex+0.5);
}
my = (int)(rmsd_mat[i][j]*delta_scaley*del_lev+0.5);
delta_tot[mx] += 1.0;
if ((rmsd_mat[i][j] >= 0) && (rmsd_mat[i][j] <= delta_maxy))
{
delta[mx][my] += 1.0;
}
}
}
}
delta_max = 0;
for (i = 0; i < delta_xsize; i++)
{
if (delta_tot[i] > 0.0)
{
delta_tot[i] = 1.0/delta_tot[i];
for (j = 0; j <= del_lev; j++)
{
delta[i][j] *= delta_tot[i];
if (delta[i][j] > delta_max)
{
delta_max = delta[i][j];
}
}
}
}
fprintf(stderr, "Maximum in delta matrix: %f\n", delta_max);
snew(del_xaxis, delta_xsize);
snew(del_yaxis, del_lev+1);
for (i = 0; i < delta_xsize; i++)
{
del_xaxis[i] = axis[i]-axis[0];
}
for (i = 0; i < del_lev+1; i++)
{
del_yaxis[i] = delta_maxy*i/del_lev;
}
sprintf(buf, "%s %s vs. delta t", gn_rms[0], whatname[ewhat]);
fp = gmx_ffopen("delta.xpm", "w");
write_xpm(fp, 0, buf, "density", output_env_get_time_label(oenv), whatlabel[ewhat],
delta_xsize, del_lev+1, del_xaxis, del_yaxis,
delta, 0.0, delta_max, rlo, rhi, &nlevels);
gmx_ffclose(fp);
}
if (opt2bSet("-bin", NFILE, fnm))
{
/* NB: File must be binary if we use fwrite */
fp = ftp2FILE(efDAT, NFILE, fnm, "wb");
for (i = 0; i < tel_mat; i++)
{
if (fwrite(rmsd_mat[i], sizeof(**rmsd_mat), tel_mat2, fp) != tel_mat2)
{
gmx_fatal(FARGS, "Error writing to output file");
}
}
gmx_ffclose(fp);
}
}
if (bBond)
{
fprintf(stderr, "\nMin. angle: %f, Max. angle: %f\n", bond_min, bond_max);
if (bond_user_max != -1)
{
bond_max = bond_user_max;
}
if (bond_user_min != -1)
{
bond_min = bond_user_min;
}
if ((bond_user_max != -1) || (bond_user_min != -1))
{
fprintf(stderr, "Bond angle Min and Max set to:\n"
"Min. angle: %f, Max. angle: %f\n", bond_min, bond_max);
}
rlo.r = 1; rlo.g = 1; rlo.b = 1;
rhi.r = 0; rhi.g = 0; rhi.b = 0;
sprintf(buf, "%s av. bond angle deviation", gn_rms[0]);
write_xpm(opt2FILE("-bm", NFILE, fnm, "w"), 0, buf, "degrees",
output_env_get_time_label(oenv), output_env_get_time_label(oenv), tel_mat, tel_mat2,
axis, axis2, bond_mat, bond_min, bond_max, rlo, rhi, &nlevels);
}
}
bAv = opt2bSet("-a", NFILE, fnm);
/* Write the RMSD's to file */
if (!bPrev)
{
sprintf(buf, "%s", whatxvgname[ewhat]);
}
else
{
sprintf(buf, "%s with frame %g %s ago", whatxvgname[ewhat],
time[prev*freq]-time[0], output_env_get_time_label(oenv));
}
fp = xvgropen(opt2fn("-o", NFILE, fnm), buf, output_env_get_xvgr_tlabel(oenv),
whatxvglabel[ewhat], oenv);
if (output_env_get_print_xvgr_codes(oenv))
{
fprintf(fp, "@ subtitle \"%s%s after %s%s%s\"\n",
(nrms == 1) ? "" : "of ", gn_rms[0], fitgraphlabel[efit],
bFit ? " to " : "", bFit ? gn_fit : "");
}
if (nrms != 1)
{
xvgr_legend(fp, nrms, (const char**)gn_rms, oenv);
}
for (i = 0; (i < teller); i++)
{
if (bSplit && i > 0 &&
abs(time[bPrev ? freq*i : i]/output_env_get_time_factor(oenv)) < 1e-5)
{
fprintf(fp, "&\n");
}
fprintf(fp, "%12.7f", time[bPrev ? freq*i : i]);
for (j = 0; (j < nrms); j++)
{
fprintf(fp, " %12.7f", rls[j][i]);
if (bAv)
{
rlstot += rls[j][i];
}
}
fprintf(fp, "\n");
}
gmx_ffclose(fp);
if (bMirror)
{
/* Write the mirror RMSD's to file */
sprintf(buf, "%s with Mirror", whatxvgname[ewhat]);
sprintf(buf2, "Mirror %s", whatxvglabel[ewhat]);
fp = xvgropen(opt2fn("-mir", NFILE, fnm), buf, output_env_get_xvgr_tlabel(oenv),
buf2, oenv);
if (nrms == 1)
{
if (output_env_get_print_xvgr_codes(oenv))
{
fprintf(fp, "@ subtitle \"of %s after lsq fit to mirror of %s\"\n",
gn_rms[0], gn_fit);
}
}
else
{
if (output_env_get_print_xvgr_codes(oenv))
{
fprintf(fp, "@ subtitle \"after lsq fit to mirror %s\"\n", gn_fit);
}
xvgr_legend(fp, nrms, (const char**)gn_rms, oenv);
}
for (i = 0; (i < teller); i++)
{
if (bSplit && i > 0 && abs(time[i]) < 1e-5)
{
fprintf(fp, "&\n");
}
fprintf(fp, "%12.7f", time[i]);
for (j = 0; (j < nrms); j++)
{
fprintf(fp, " %12.7f", rlsm[j][i]);
}
fprintf(fp, "\n");
}
gmx_ffclose(fp);
}
if (bAv)
{
sprintf(buf, "Average %s", whatxvgname[ewhat]);
sprintf(buf2, "Average %s", whatxvglabel[ewhat]);
fp = xvgropen(opt2fn("-a", NFILE, fnm), buf, "Residue", buf2, oenv);
for (j = 0; (j < nrms); j++)
{
fprintf(fp, "%10d %10g\n", j, rlstot/teller);
}
gmx_ffclose(fp);
}
if (bNorm)
{
fp = xvgropen("aver.xvg", gn_rms[0], "Residue", whatxvglabel[ewhat], oenv);
for (j = 0; (j < irms[0]); j++)
{
fprintf(fp, "%10d %10g\n", j, rlsnorm[j]/teller);
}
gmx_ffclose(fp);
}
do_view(oenv, opt2fn_null("-a", NFILE, fnm), "-graphtype bar");
do_view(oenv, opt2fn("-o", NFILE, fnm), NULL);
do_view(oenv, opt2fn_null("-mir", NFILE, fnm), NULL);
do_view(oenv, opt2fn_null("-m", NFILE, fnm), NULL);
do_view(oenv, opt2fn_null("-bm", NFILE, fnm), NULL);
do_view(oenv, opt2fn_null("-dist", NFILE, fnm), NULL);
return 0;
}
SEXP Riproc_recv_model(SEXP ytime, SEXP ysender, SEXP yreceiver,
SEXP xtime, SEXP xsender, SEXP xreceiver,
SEXP factors, SEXP variables, SEXP nsend, SEXP nrecv,
SEXP loops)
{
struct properties p;
struct history hx, hy;
struct design s, r;
struct design2 d;
struct terms_object tm;
const struct message *msgs;
size_t nmsg;
size_t iter, dim, nc, ntot, rank;
size_t ncextra;
struct recv_fit fit;
const struct recv_loglik *ll;
const struct recv_model *model;
const struct constr *constr;
const double *beta, *nu;
double dev;
SEXP names;
SEXP ret = NULL_USER_OBJECT;
int k;
int err = 0;
err = get_properties(nsend, nrecv, loops, &p);
if (err < 0)
goto properties_fail;
history_init(&hy, p.nsend, p.nrecv);
err = get_history(ytime, ysender, yreceiver, &hy);
if (err < 0)
goto yhistory_fail;
history_init(&hx, p.nsend, p.nrecv);
err = get_history(xtime, xsender, xreceiver, &hx);
if (err < 0)
goto xhistory_fail;
design_init(&s, &hx, p.nsend);
design_init(&r, &hx, p.nrecv);
design2_init(&d, &hx, p.nsend, p.nrecv);
err = get_terms_object(factors, variables, &s, &r, &d, &tm);
if (err < 0)
goto terms_fail;
history_get_messages(&hy, &msgs, &nmsg);
recv_fit_init(&fit, &r, &d, p.exclude_loops, msgs, nmsg, NULL, NULL);
ncextra = recv_fit_extra_constr_count(&fit);
if (ncextra)
warning("adding %zd %s to make parameters identifiable\n",
ncextra, ncextra == 1 ? "constraint" : "constraints");
err = do_fit(&fit, NULL, NULL, &iter);
if (err < 0)
goto fit_fail;
constr = recv_fit_constr(&fit);
nc = constr_count(constr);
ll = recv_fit_loglik(&fit);
ntot = recv_loglik_count(ll);
model = recv_loglik_model(ll);
beta = (recv_model_params(model))->recv.traits; /* HACK */
nu = recv_fit_duals(&fit);
dim = recv_model_dim(model);
rank = dim - nc;
dev = recv_loglik_dev(ll);
PROTECT(ret = NEW_LIST(16));
PROTECT(names = NEW_CHARACTER(16));
SET_NAMES(ret, names);
k = 0;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("coefficients"));
SET_VECTOR_ELT(ret, k, alloc_vector_copy(beta, dim));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("duals"));
SET_VECTOR_ELT(ret, k, alloc_vector_copy(nu, nc));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("constraints"));
SET_VECTOR_ELT(ret, k, alloc_matrix_copy(constr_all_wts(constr), nc, dim));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("constraint.values"));
SET_VECTOR_ELT(ret, k, alloc_vector_copy(constr_all_vals(constr), nc));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("rank"));
SET_VECTOR_ELT(ret, k, ScalarInteger((int)rank));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("deviance"));
SET_VECTOR_ELT(ret, k, ScalarReal(dev));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("aic"));
SET_VECTOR_ELT(ret, k, ScalarReal(dev + (double) 2 * rank));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("null.deviance"));
SET_VECTOR_ELT(ret, k, ScalarReal(recv_fit_dev0(&fit)));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("iter"));
SET_VECTOR_ELT(ret, k, ScalarInteger((int)iter));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("df.residual"));
SET_VECTOR_ELT(ret, k, ScalarReal((double)(ntot - rank)));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("df.null"));
SET_VECTOR_ELT(ret, k, ScalarReal((double)ntot));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("converged"));
SET_VECTOR_ELT(ret, k, ScalarLogical(TRUE));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("names"));
SET_VECTOR_ELT(ret, k, alloc_term_labels(&tm.terms));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("perm"));
SET_VECTOR_ELT(ret, k, alloc_terms_permute(&tm.terms, &r, &d));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("score"));
SET_VECTOR_ELT(ret, k, alloc_score(ll));
k++;
SET_STRING_ELT(names, k, COPY_TO_USER_STRING("imat"));
SET_VECTOR_ELT(ret, k, alloc_imat(&fit));
k++;
UNPROTECT(2);
fit_fail:
recv_fit_deinit(&fit);
terms_fail:
design2_deinit(&d);
design_deinit(&r);
design_deinit(&s);
xhistory_fail:
history_deinit(&hx);
yhistory_fail:
history_deinit(&hy);
properties_fail:
return ret;
}
real fit_ahx(int nres,t_bb bb[],int natoms,int nall,atom_id allindex[],
rvec x[],int nca,
atom_id caindex[],matrix box,bool bFit)
{
static rvec *xref=NULL;
static real *mass=NULL;
const real d=0.15; /* Rise per residue (nm) */
const real tw=1.745; /* Twist per residue (rad) */
const real rad=0.23; /* Radius of the helix (nm) */
real phi0,trms,rms;
rvec dx,xcm;
int ai,i,nmass;
if (nca < 3)
gmx_fatal(FARGS,"Need at least 3 Calphas to fit to, (now %d)...\n",nca);
if (xref == NULL) {
snew(xref,natoms);
snew(mass,natoms);
}
phi0=0;
for(i=0; (i<nca); i++) {
ai=caindex[i];
xref[ai][XX]=rad*cos(phi0);
xref[ai][YY]=rad*sin(phi0);
xref[ai][ZZ]=d*i;
/* Set the mass to select that this Calpha contributes to fitting */
mass[ai]=10.0;
/*#define DEBUG*/
#ifdef DEBUG
fprintf(stderr,"%5d %8.3f %8.3f %8.3f %8.3f %8.3f %8.3f\n",ai,
x[ai][XX],x[ai][YY],x[ai][ZZ],
xref[ai][XX],xref[ai][YY],xref[ai][ZZ]);
#endif
phi0+=tw;
}
/* Center the referece around the origin */
my_calc_xcm(nca,caindex,xref,xcm);
my_sub_xcm(nca,caindex,xref,xcm);
if (bFit) {
/* Now center the to-be-fitted coords around the origin */
my_calc_xcm(nca,caindex,x,xcm);
my_sub_xcm(nall,allindex,x,xcm);
}
#ifdef DEBUG
dump_ahx(nres,bb,xref,box,0);
#endif
nmass=0;
for(i=0; (i<natoms); i++)
if (mass[i] > 0)
nmass++;
if (nmass != nca)
gmx_fatal(FARGS,"nmass=%d, nca=%d\n",nmass,nca);
/* Now call the fitting routine */
if (bFit)
do_fit(natoms,mass,xref,x);
/* Reset masses and calc rms */
trms=0.0;
for(i=0; (i<nres); i++) {
ai=bb[i].CA;
if (mass[ai] > 0.0) {
rvec_sub(x[ai],xref[ai],dx);
rms=iprod(dx,dx);
bb[i].rmsa+=sqrt(rms);
bb[i].nrms++;
trms+=rms;
mass[ai]=0.0;
}
}
return sqrt(trms/nca);
}
int gmx_disre(int argc,char *argv[])
{
const char *desc[] = {
"g_disre computes violations of distance restraints.",
"If necessary all protons can be added to a protein molecule ",
"using the protonate program.[PAR]",
"The program always",
"computes the instantaneous violations rather than time-averaged,",
"because this analysis is done from a trajectory file afterwards",
"it does not make sense to use time averaging. However,",
"the time averaged values per restraint are given in the log file.[PAR]",
"An index file may be used to select specific restraints for",
"printing.[PAR]",
"When the optional[TT]-q[tt] flag is given a pdb file coloured by the",
"amount of average violations.[PAR]",
"When the [TT]-c[tt] option is given, an index file will be read",
"containing the frames in your trajectory corresponding to the clusters",
"(defined in another manner) that you want to analyze. For these clusters",
"the program will compute average violations using the third power",
"averaging algorithm and print them in the log file."
};
static int ntop = 0;
static int nlevels = 20;
static real max_dr = 0;
static gmx_bool bThird = TRUE;
t_pargs pa[] = {
{ "-ntop", FALSE, etINT, {&ntop},
"Number of large violations that are stored in the log file every step" },
{ "-maxdr", FALSE, etREAL, {&max_dr},
"Maximum distance violation in matrix output. If less than or equal to 0 the maximum will be determined by the data." },
{ "-nlevels", FALSE, etINT, {&nlevels},
"Number of levels in the matrix output" },
{ "-third", FALSE, etBOOL, {&bThird},
"Use inverse third power averaging or linear for matrix output" }
};
FILE *out=NULL,*aver=NULL,*numv=NULL,*maxxv=NULL,*xvg=NULL;
t_tpxheader header;
t_inputrec ir;
gmx_mtop_t mtop;
rvec *xtop;
gmx_localtop_t *top;
t_atoms *atoms=NULL;
t_forcerec *fr;
t_fcdata fcd;
t_nrnb nrnb;
t_commrec *cr;
t_graph *g;
int ntopatoms,natoms,i,j,kkk;
t_trxstatus *status;
real t;
rvec *x,*f,*xav=NULL;
matrix box;
gmx_bool bPDB;
int isize;
atom_id *index=NULL,*ind_fit=NULL;
char *grpname;
t_cluster_ndx *clust=NULL;
t_dr_result dr,*dr_clust=NULL;
char **leg;
real *vvindex=NULL,*w_rls=NULL;
t_mdatoms *mdatoms;
t_pbc pbc,*pbc_null;
int my_clust;
FILE *fplog;
output_env_t oenv;
gmx_rmpbc_t gpbc=NULL;
t_filenm fnm[] = {
{ efTPX, NULL, NULL, ffREAD },
{ efTRX, "-f", NULL, ffREAD },
{ efXVG, "-ds", "drsum", ffWRITE },
{ efXVG, "-da", "draver", ffWRITE },
{ efXVG, "-dn", "drnum", ffWRITE },
{ efXVG, "-dm", "drmax", ffWRITE },
{ efXVG, "-dr", "restr", ffWRITE },
{ efLOG, "-l", "disres", ffWRITE },
{ efNDX, NULL, "viol", ffOPTRD },
{ efPDB, "-q", "viol", ffOPTWR },
{ efNDX, "-c", "clust", ffOPTRD },
{ efXPM, "-x", "matrix", ffOPTWR }
};
#define NFILE asize(fnm)
cr = init_par(&argc,&argv);
CopyRight(stderr,argv[0]);
parse_common_args(&argc,argv,PCA_CAN_TIME | PCA_CAN_VIEW | PCA_BE_NICE,
NFILE,fnm,asize(pa),pa,asize(desc),desc,0,NULL,&oenv);
gmx_log_open(ftp2fn(efLOG,NFILE,fnm),cr,FALSE,0,&fplog);
if (ntop)
init5(ntop);
read_tpxheader(ftp2fn(efTPX,NFILE,fnm),&header,FALSE,NULL,NULL);
snew(xtop,header.natoms);
read_tpx(ftp2fn(efTPX,NFILE,fnm),&ir,box,&ntopatoms,xtop,NULL,NULL,&mtop);
bPDB = opt2bSet("-q",NFILE,fnm);
if (bPDB) {
snew(xav,ntopatoms);
snew(ind_fit,ntopatoms);
snew(w_rls,ntopatoms);
for(kkk=0; (kkk<ntopatoms); kkk++) {
w_rls[kkk] = 1;
ind_fit[kkk] = kkk;
}
snew(atoms,1);
*atoms = gmx_mtop_global_atoms(&mtop);
if (atoms->pdbinfo == NULL) {
snew(atoms->pdbinfo,atoms->nr);
}
}
top = gmx_mtop_generate_local_top(&mtop,&ir);
g = NULL;
pbc_null = NULL;
if (ir.ePBC != epbcNONE) {
if (ir.bPeriodicMols)
pbc_null = &pbc;
else
g = mk_graph(fplog,&top->idef,0,mtop.natoms,FALSE,FALSE);
}
if (ftp2bSet(efNDX,NFILE,fnm)) {
rd_index(ftp2fn(efNDX,NFILE,fnm),1,&isize,&index,&grpname);
xvg=xvgropen(opt2fn("-dr",NFILE,fnm),"Inidividual Restraints","Time (ps)",
"nm",oenv);
snew(vvindex,isize);
snew(leg,isize);
for(i=0; (i<isize); i++) {
index[i]++;
snew(leg[i],12);
sprintf(leg[i],"index %d",index[i]);
}
xvgr_legend(xvg,isize,(const char**)leg,oenv);
}
else
isize=0;
ir.dr_tau=0.0;
init_disres(fplog,&mtop,&ir,NULL,FALSE,&fcd,NULL);
natoms=read_first_x(oenv,&status,ftp2fn(efTRX,NFILE,fnm),&t,&x,box);
snew(f,5*natoms);
init_dr_res(&dr,fcd.disres.nres);
if (opt2bSet("-c",NFILE,fnm)) {
clust = cluster_index(fplog,opt2fn("-c",NFILE,fnm));
snew(dr_clust,clust->clust->nr+1);
for(i=0; (i<=clust->clust->nr); i++)
init_dr_res(&dr_clust[i],fcd.disres.nres);
}
else {
out =xvgropen(opt2fn("-ds",NFILE,fnm),
"Sum of Violations","Time (ps)","nm",oenv);
aver=xvgropen(opt2fn("-da",NFILE,fnm),
"Average Violation","Time (ps)","nm",oenv);
numv=xvgropen(opt2fn("-dn",NFILE,fnm),
"# Violations","Time (ps)","#",oenv);
maxxv=xvgropen(opt2fn("-dm",NFILE,fnm),
"Largest Violation","Time (ps)","nm",oenv);
}
mdatoms = init_mdatoms(fplog,&mtop,ir.efep!=efepNO);
atoms2md(&mtop,&ir,0,NULL,0,mtop.natoms,mdatoms);
update_mdatoms(mdatoms,ir.init_lambda);
fr = mk_forcerec();
fprintf(fplog,"Made forcerec\n");
init_forcerec(fplog,oenv,fr,NULL,&ir,&mtop,cr,box,FALSE,NULL,NULL,NULL,
FALSE,-1);
init_nrnb(&nrnb);
if (ir.ePBC != epbcNONE)
gpbc = gmx_rmpbc_init(&top->idef,ir.ePBC,natoms,box);
j=0;
do {
if (ir.ePBC != epbcNONE) {
if (ir.bPeriodicMols)
set_pbc(&pbc,ir.ePBC,box);
else
gmx_rmpbc(gpbc,natoms,box,x);
}
if (clust) {
if (j > clust->maxframe)
gmx_fatal(FARGS,"There are more frames in the trajectory than in the cluster index file. t = %8f\n",t);
my_clust = clust->inv_clust[j];
range_check(my_clust,0,clust->clust->nr);
check_viol(fplog,cr,&(top->idef.il[F_DISRES]),
top->idef.iparams,top->idef.functype,
x,f,fr,pbc_null,g,dr_clust,my_clust,isize,index,vvindex,&fcd);
}
else
check_viol(fplog,cr,&(top->idef.il[F_DISRES]),
top->idef.iparams,top->idef.functype,
x,f,fr,pbc_null,g,&dr,0,isize,index,vvindex,&fcd);
if (bPDB) {
reset_x(atoms->nr,ind_fit,atoms->nr,NULL,x,w_rls);
do_fit(atoms->nr,w_rls,x,x);
if (j == 0) {
/* Store the first frame of the trajectory as 'characteristic'
* for colouring with violations.
*/
for(kkk=0; (kkk<atoms->nr); kkk++)
copy_rvec(x[kkk],xav[kkk]);
}
}
if (!clust) {
if (isize > 0) {
fprintf(xvg,"%10g",t);
for(i=0; (i<isize); i++)
fprintf(xvg," %10g",vvindex[i]);
fprintf(xvg,"\n");
}
fprintf(out, "%10g %10g\n",t,dr.sumv);
fprintf(aver, "%10g %10g\n",t,dr.averv);
fprintf(maxxv,"%10g %10g\n",t,dr.maxv);
fprintf(numv, "%10g %10d\n",t,dr.nv);
}
j++;
} while (read_next_x(oenv,status,&t,natoms,x,box));
close_trj(status);
if (ir.ePBC != epbcNONE)
gmx_rmpbc_done(gpbc);
if (clust) {
dump_clust_stats(fplog,fcd.disres.nres,&(top->idef.il[F_DISRES]),
top->idef.iparams,clust->clust,dr_clust,
clust->grpname,isize,index);
}
else {
dump_stats(fplog,j,fcd.disres.nres,&(top->idef.il[F_DISRES]),
top->idef.iparams,&dr,isize,index,
bPDB ? atoms : NULL);
if (bPDB) {
write_sto_conf(opt2fn("-q",NFILE,fnm),
"Coloured by average violation in Angstrom",
atoms,xav,NULL,ir.ePBC,box);
}
dump_disre_matrix(opt2fn_null("-x",NFILE,fnm),&dr,fcd.disres.nres,
j,&top->idef,&mtop,max_dr,nlevels,bThird);
ffclose(out);
ffclose(aver);
ffclose(numv);
ffclose(maxxv);
if (isize > 0) {
ffclose(xvg);
do_view(oenv,opt2fn("-dr",NFILE,fnm),"-nxy");
}
do_view(oenv,opt2fn("-dn",NFILE,fnm),"-nxy");
do_view(oenv,opt2fn("-da",NFILE,fnm),"-nxy");
do_view(oenv,opt2fn("-ds",NFILE,fnm),"-nxy");
do_view(oenv,opt2fn("-dm",NFILE,fnm),"-nxy");
}
thanx(stderr);
if (gmx_parallel_env_initialized())
gmx_finalize();
gmx_log_close(fplog);
return 0;
}
static void analyze_clusters(int nf, t_clusters *clust, real **rmsd,
int natom, t_atoms *atoms, rvec *xtps,
real *mass, rvec **xx, real *time,
int ifsize, atom_id *fitidx,
int iosize, atom_id *outidx,
char *trxfn, char *sizefn, char *transfn,
char *ntransfn, char *clustidfn, bool bAverage,
int write_ncl, int write_nst, real rmsmin,bool bFit,
FILE *log,t_rgb rlo,t_rgb rhi)
{
FILE *fp=NULL;
char buf[STRLEN],buf1[40],buf2[40],buf3[40],*trxsfn;
int trxout=0,trxsout=0;
int i,i1,cl,nstr,*structure,first=0,midstr;
bool *bWrite=NULL;
real r,clrmsd,midrmsd;
rvec *xav=NULL;
matrix zerobox;
clear_mat(zerobox);
ffprintf1(stderr,log,buf,"\nFound %d clusters\n\n",clust->ncl);
trxsfn=NULL;
if (trxfn) {
/* do we write all structures? */
if (write_ncl) {
trxsfn = parse_filename(trxfn, max(write_ncl,clust->ncl));
snew(bWrite,nf);
}
ffprintf2(stderr,log,buf,"Writing %s structure for each cluster to %s\n",
bAverage ? "average" : "middle", trxfn);
if (write_ncl) {
/* find out what we want to tell the user:
Writing [all structures|structures with rmsd > %g] for
{all|first %d} clusters {with more than %d structures} to %s */
if (rmsmin>0.0)
sprintf(buf1,"structures with rmsd > %g",rmsmin);
else
sprintf(buf1,"all structures");
buf2[0]=buf3[0]='\0';
if (write_ncl>=clust->ncl) {
if (write_nst==0)
sprintf(buf2,"all ");
} else
sprintf(buf2,"the first %d ",write_ncl);
if (write_nst)
sprintf(buf3," with more than %d structures",write_nst);
sprintf(buf,"Writing %s for %sclusters%s to %s\n",buf1,buf2,buf3,trxsfn);
ffprintf(stderr,log,buf);
}
/* Prepare a reference structure for the orientation of the clusters */
if (bFit)
reset_x(ifsize,fitidx,natom,NULL,xtps,mass);
trxout = open_trx(trxfn,"w");
/* Calculate the average structure in each cluster, *
* all structures are fitted to the first struture of the cluster */
snew(xav,natom);
}
if (transfn || ntransfn)
ana_trans(clust, nf, transfn, ntransfn, log,rlo,rhi);
if (clustidfn) {
fp=xvgropen(clustidfn,"Clusters",xvgr_tlabel(),"Cluster #");
fprintf(fp,"@ s0 symbol 2\n");
fprintf(fp,"@ s0 symbol size 0.2\n");
fprintf(fp,"@ s0 linestyle 0\n");
for(i=0; i<nf; i++)
fprintf(fp,"%8g %8d\n",time[i],clust->cl[i]);
ffclose(fp);
}
if (sizefn) {
fp=xvgropen(sizefn,"Cluster Sizes","Cluster #","# Structures");
fprintf(fp,"@g%d type %s\n",0,"bar");
}
snew(structure,nf);
fprintf(log,"\n%3s | %3s %4s | %6s %4s | cluster members\n",
"cl.","#st","rmsd","middle","rmsd");
for(cl=1; cl<=clust->ncl; cl++) {
/* prepare structures (fit, middle, average) */
if (xav)
for(i=0; i<natom;i++)
clear_rvec(xav[i]);
nstr=0;
for(i1=0; i1<nf; i1++)
if (clust->cl[i1] == cl) {
structure[nstr] = i1;
nstr++;
if (trxfn && (bAverage || write_ncl) ) {
if (bFit)
reset_x(ifsize,fitidx,natom,NULL,xx[i1],mass);
if (nstr == 1)
first = i1;
else if (bFit)
do_fit(natom,mass,xx[first],xx[i1]);
if (xav)
for(i=0; i<natom; i++)
rvec_inc(xav[i],xx[i1][i]);
}
}
if (sizefn)
fprintf(fp,"%8d %8d\n",cl,nstr);
clrmsd = 0;
midstr = 0;
midrmsd = 10000;
for(i1=0; i1<nstr; i1++) {
r = 0;
if (nstr > 1) {
for(i=0; i<nstr; i++)
if (i < i1)
r += rmsd[structure[i]][structure[i1]];
else
r += rmsd[structure[i1]][structure[i]];
r /= (nstr - 1);
}
if ( r < midrmsd ) {
midstr = structure[i1];
midrmsd = r;
}
clrmsd += r;
}
clrmsd /= nstr;
/* dump cluster info to logfile */
if (nstr > 1) {
sprintf(buf1,"%5.3f",clrmsd);
if (buf1[0] == '0')
buf1[0] = ' ';
sprintf(buf2,"%5.3f",midrmsd);
if (buf2[0] == '0')
buf2[0] = ' ';
} else {
sprintf(buf1,"%5s","");
sprintf(buf2,"%5s","");
}
fprintf(log,"%3d | %3d%s | %6g%s |",cl,nstr,buf1,time[midstr],buf2);
for(i=0; i<nstr; i++) {
if ((i % 7 == 0) && i)
sprintf(buf,"\n%3s | %3s %4s | %6s %4s |","","","","","");
else
buf[0] = '\0';
i1 = structure[i];
fprintf(log,"%s %6g",buf,time[i1]);
}
fprintf(log,"\n");
/* write structures to trajectory file(s) */
if (trxfn) {
if (write_ncl)
for(i=0; i<nstr; i++)
bWrite[i]=FALSE;
if ( cl < write_ncl+1 && nstr > write_nst ) {
/* Dump all structures for this cluster */
/* generate numbered filename (there is a %d in trxfn!) */
sprintf(buf,trxsfn,cl);
trxsout = open_trx(buf,"w");
for(i=0; i<nstr; i++) {
bWrite[i] = TRUE;
if (rmsmin>0.0)
for(i1=0; i1<i && bWrite[i]; i1++)
if (bWrite[i1])
bWrite[i] = rmsd[structure[i1]][structure[i]] > rmsmin;
if (bWrite[i])
write_trx(trxsout,iosize,outidx,atoms,i,time[structure[i]],zerobox,
xx[structure[i]],NULL);
}
close_trx(trxsout);
}
/* Dump the average structure for this cluster */
if (bAverage) {
for(i=0; i<natom; i++)
svmul(1.0/nstr,xav[i],xav[i]);
} else {
for(i=0; i<natom; i++)
copy_rvec(xx[midstr][i],xav[i]);
if (bFit)
reset_x(ifsize,fitidx,natom,NULL,xav,mass);
}
if (bFit)
do_fit(natom,mass,xtps,xav);
r = cl;
write_trx(trxout,iosize,outidx,atoms,cl,time[midstr],zerobox,xav,NULL);
}
}
/* clean up */
if (trxfn) {
close_trx(trxout);
sfree(xav);
if (write_ncl)
sfree(bWrite);
}
sfree(structure);
if (trxsfn)
sfree(trxsfn);
}
int gmx_analyze(int argc,char *argv[])
{
static const char *desc[] = {
"g_analyze reads an ascii file and analyzes data sets.",
"A line in the input file may start with a time",
"(see option [TT]-time[tt]) and any number of y values may follow.",
"Multiple sets can also be",
"read when they are seperated by & (option [TT]-n[tt]),",
"in this case only one y value is read from each line.",
"All lines starting with # and @ are skipped.",
"All analyses can also be done for the derivative of a set",
"(option [TT]-d[tt]).[PAR]",
"All options, except for [TT]-av[tt] and [TT]-power[tt] assume that the",
"points are equidistant in time.[PAR]",
"g_analyze always shows the average and standard deviation of each",
"set. For each set it also shows the relative deviation of the third",
"and forth cumulant from those of a Gaussian distribution with the same",
"standard deviation.[PAR]",
"Option [TT]-ac[tt] produces the autocorrelation function(s).[PAR]",
"Option [TT]-cc[tt] plots the resemblance of set i with a cosine of",
"i/2 periods. The formula is:[BR]"
"2 (int0-T y(t) cos(i pi t) dt)^2 / int0-T y(t) y(t) dt[BR]",
"This is useful for principal components obtained from covariance",
"analysis, since the principal components of random diffusion are",
"pure cosines.[PAR]",
"Option [TT]-msd[tt] produces the mean square displacement(s).[PAR]",
"Option [TT]-dist[tt] produces distribution plot(s).[PAR]",
"Option [TT]-av[tt] produces the average over the sets.",
"Error bars can be added with the option [TT]-errbar[tt].",
"The errorbars can represent the standard deviation, the error",
"(assuming the points are independent) or the interval containing",
"90% of the points, by discarding 5% of the points at the top and",
"the bottom.[PAR]",
"Option [TT]-ee[tt] produces error estimates using block averaging.",
"A set is divided in a number of blocks and averages are calculated for",
"each block. The error for the total average is calculated from",
"the variance between averages of the m blocks B_i as follows:",
"error^2 = Sum (B_i - <B>)^2 / (m*(m-1)).",
"These errors are plotted as a function of the block size.",
"Also an analytical block average curve is plotted, assuming",
"that the autocorrelation is a sum of two exponentials.",
"The analytical curve for the block average is:[BR]",
"f(t) = sigma sqrt(2/T ( a (tau1 ((exp(-t/tau1) - 1) tau1/t + 1)) +[BR]",
" (1-a) (tau2 ((exp(-t/tau2) - 1) tau2/t + 1)))),[BR]"
"where T is the total time.",
"a, tau1 and tau2 are obtained by fitting f^2(t) to error^2.",
"When the actual block average is very close to the analytical curve,",
"the error is sigma*sqrt(2/T (a tau1 + (1-a) tau2)).",
"The complete derivation is given in",
"B. Hess, J. Chem. Phys. 116:209-217, 2002.[PAR]",
"Option [TT]-filter[tt] prints the RMS high-frequency fluctuation",
"of each set and over all sets with respect to a filtered average.",
"The filter is proportional to cos(pi t/len) where t goes from -len/2",
"to len/2. len is supplied with the option [TT]-filter[tt].",
"This filter reduces oscillations with period len/2 and len by a factor",
"of 0.79 and 0.33 respectively.[PAR]",
"Option [TT]-g[tt] fits the data to the function given with option",
"[TT]-fitfn[tt].[PAR]",
"Option [TT]-power[tt] fits the data to b t^a, which is accomplished",
"by fitting to a t + b on log-log scale. All points after the first",
"zero or negative value are ignored.[PAR]"
"Option [TT]-luzar[tt] performs a Luzar & Chandler kinetics analysis",
"on output from [TT]g_hbond[tt]. The input file can be taken directly",
"from [TT]g_hbond -ac[tt], and then the same result should be produced."
};
static real tb=-1,te=-1,frac=0.5,filtlen=0,binwidth=0.1,aver_start=0;
static bool bHaveT=TRUE,bDer=FALSE,bSubAv=TRUE,bAverCorr=FALSE,bXYdy=FALSE;
static bool bEESEF=FALSE,bEENLC=FALSE,bEeFitAc=FALSE,bPower=FALSE;
static bool bIntegrate=FALSE,bRegression=FALSE,bLuzar=FALSE,bLuzarError=FALSE;
static int nsets_in=1,d=1,nb_min=4,resol=10;
static real temp=298.15,fit_start=1,smooth_tail_start=-1;
/* must correspond to enum avbar* declared at beginning of file */
static const char *avbar_opt[avbarNR+1] = {
NULL, "none", "stddev", "error", "90", NULL
};
t_pargs pa[] = {
{ "-time", FALSE, etBOOL, {&bHaveT},
"Expect a time in the input" },
{ "-b", FALSE, etREAL, {&tb},
"First time to read from set" },
{ "-e", FALSE, etREAL, {&te},
"Last time to read from set" },
{ "-n", FALSE, etINT, {&nsets_in},
"Read # sets seperated by &" },
{ "-d", FALSE, etBOOL, {&bDer},
"Use the derivative" },
{ "-dp", FALSE, etINT, {&d},
"HIDDENThe derivative is the difference over # points" },
{ "-bw", FALSE, etREAL, {&binwidth},
"Binwidth for the distribution" },
{ "-errbar", FALSE, etENUM, {avbar_opt},
"Error bars for -av" },
{ "-integrate",FALSE,etBOOL, {&bIntegrate},
"Integrate data function(s) numerically using trapezium rule" },
{ "-aver_start",FALSE, etREAL, {&aver_start},
"Start averaging the integral from here" },
{ "-xydy", FALSE, etBOOL, {&bXYdy},
"Interpret second data set as error in the y values for integrating" },
{ "-regression",FALSE,etBOOL,{&bRegression},
"Perform a linear regression analysis on the data" },
{ "-luzar", FALSE, etBOOL, {&bLuzar},
"Do a Luzar and Chandler analysis on a correlation function and related as produced by g_hbond. When in addition the -xydy flag is given the second and fourth column will be interpreted as errors in c(t) and n(t)." },
{ "-temp", FALSE, etREAL, {&temp},
"Temperature for the Luzar hydrogen bonding kinetics analysis" },
{ "-fitstart", FALSE, etREAL, {&fit_start},
"Time (ps) from which to start fitting the correlation functions in order to obtain the forward and backward rate constants for HB breaking and formation" },
{ "-smooth",FALSE, etREAL, {&smooth_tail_start},
"If >= 0, the tail of the ACF will be smoothed by fitting it to an exponential function: y = A exp(-x/tau)" },
{ "-nbmin", FALSE, etINT, {&nb_min},
"HIDDENMinimum number of blocks for block averaging" },
{ "-resol", FALSE, etINT, {&resol},
"HIDDENResolution for the block averaging, block size increases with"
" a factor 2^(1/#)" },
{ "-eeexpfit", FALSE, etBOOL, {&bEESEF},
"HIDDENAlways use a single exponential fit for the error estimate" },
{ "-eenlc", FALSE, etBOOL, {&bEENLC},
"HIDDENAllow a negative long-time correlation" },
{ "-eefitac", FALSE, etBOOL, {&bEeFitAc},
"HIDDENAlso plot analytical block average using a autocorrelation fit" },
{ "-filter", FALSE, etREAL, {&filtlen},
"Print the high-frequency fluctuation after filtering with a cosine filter of length #" },
{ "-power", FALSE, etBOOL, {&bPower},
"Fit data to: b t^a" },
{ "-subav", FALSE, etBOOL, {&bSubAv},
"Subtract the average before autocorrelating" },
{ "-oneacf", FALSE, etBOOL, {&bAverCorr},
"Calculate one ACF over all sets" }
};
#define NPA asize(pa)
FILE *out,*out_fit;
int n,nlast,s,nset,i,j=0;
real **val,*t,dt,tot,error;
double *av,*sig,cum1,cum2,cum3,cum4,db;
char *acfile,*msdfile,*ccfile,*distfile,*avfile,*eefile,*fitfile;
t_filenm fnm[] = {
{ efXVG, "-f", "graph", ffREAD },
{ efXVG, "-ac", "autocorr", ffOPTWR },
{ efXVG, "-msd", "msd", ffOPTWR },
{ efXVG, "-cc", "coscont", ffOPTWR },
{ efXVG, "-dist", "distr", ffOPTWR },
{ efXVG, "-av", "average", ffOPTWR },
{ efXVG, "-ee", "errest", ffOPTWR },
{ efLOG, "-g", "fitlog", ffOPTWR }
};
#define NFILE asize(fnm)
int npargs;
t_pargs *ppa;
npargs = asize(pa);
ppa = add_acf_pargs(&npargs,pa);
CopyRight(stderr,argv[0]);
parse_common_args(&argc,argv,PCA_CAN_VIEW,
NFILE,fnm,npargs,ppa,asize(desc),desc,0,NULL);
acfile = opt2fn_null("-ac",NFILE,fnm);
msdfile = opt2fn_null("-msd",NFILE,fnm);
ccfile = opt2fn_null("-cc",NFILE,fnm);
distfile = opt2fn_null("-dist",NFILE,fnm);
avfile = opt2fn_null("-av",NFILE,fnm);
eefile = opt2fn_null("-ee",NFILE,fnm);
if (opt2parg_bSet("-fitfn",npargs,ppa))
fitfile = opt2fn("-g",NFILE,fnm);
else
fitfile = opt2fn_null("-g",NFILE,fnm);
val=read_xvg_time(opt2fn("-f",NFILE,fnm),bHaveT,
opt2parg_bSet("-b",npargs,ppa),tb,
opt2parg_bSet("-e",npargs,ppa),te,
nsets_in,&nset,&n,&dt,&t);
printf("Read %d sets of %d points, dt = %g\n\n",nset,n,dt);
if (bDer) {
printf("Calculating the derivative as (f[i+%d]-f[i])/(%d*dt)\n\n",
d,d);
n -= d;
for(s=0; s<nset; s++)
for(i=0; (i<n); i++)
val[s][i] = (val[s][i+d]-val[s][i])/(d*dt);
}
if (bIntegrate) {
real sum,stddev;
printf("Calculating the integral using the trapezium rule\n");
if (bXYdy) {
sum = evaluate_integral(n,t,val[0],val[1],aver_start,&stddev);
printf("Integral %10.3f +/- %10.5f\n",sum,stddev);
}
else {
for(s=0; s<nset; s++) {
sum = evaluate_integral(n,t,val[s],NULL,aver_start,&stddev);
printf("Integral %d %10.5f +/- %10.5f\n",s+1,sum,stddev);
}
}
}
if (fitfile) {
out_fit = ffopen(fitfile,"w");
if (bXYdy && nset>=2) {
do_fit(out_fit,0,TRUE,n,t,val,npargs,ppa);
} else {
for(s=0; s<nset; s++)
do_fit(out_fit,s,FALSE,n,t,val,npargs,ppa);
}
fclose(out_fit);
}
printf(" std. dev. relative deviation of\n");
printf(" standard --------- cumulants from those of\n");
printf("set average deviation sqrt(n-1) a Gaussian distribition\n");
printf(" cum. 3 cum. 4\n");
snew(av,nset);
snew(sig,nset);
for(s=0; (s<nset); s++) {
cum1 = 0;
cum2 = 0;
cum3 = 0;
cum4 = 0;
for(i=0; (i<n); i++)
cum1 += val[s][i];
cum1 /= n;
for(i=0; (i<n); i++) {
db = val[s][i]-cum1;
cum2 += db*db;
cum3 += db*db*db;
cum4 += db*db*db*db;
}
cum2 /= n;
cum3 /= n;
cum4 /= n;
av[s] = cum1;
sig[s] = sqrt(cum2);
if (n > 1)
error = sqrt(cum2/(n-1));
else
error = 0;
printf("SS%d %13.6e %12.6e %12.6e %6.3f %6.3f\n",
s+1,av[s],sig[s],error,
sig[s] ? cum3/(sig[s]*sig[s]*sig[s]*sqrt(8/M_PI)) : 0,
sig[s] ? cum4/(sig[s]*sig[s]*sig[s]*sig[s]*3)-1 : 0);
}
printf("\n");
if (filtlen)
filter(filtlen,n,nset,val,dt);
if (msdfile) {
out=xvgropen(msdfile,"Mean square displacement",
"time","MSD (nm\\S2\\N)");
nlast = (int)(n*frac);
for(s=0; s<nset; s++) {
for(j=0; j<=nlast; j++) {
if (j % 100 == 0)
fprintf(stderr,"\r%d",j);
tot=0;
for(i=0; i<n-j; i++)
tot += sqr(val[s][i]-val[s][i+j]);
tot /= (real)(n-j);
fprintf(out," %g %8g\n",dt*j,tot);
}
if (s<nset-1)
fprintf(out,"&\n");
}
fclose(out);
fprintf(stderr,"\r%d, time=%g\n",j-1,(j-1)*dt);
}
if (ccfile)
plot_coscont(ccfile,n,nset,val);
if (distfile)
histogram(distfile,binwidth,n,nset,val);
if (avfile)
average(avfile,nenum(avbar_opt),n,nset,val,t);
if (eefile)
estimate_error(eefile,nb_min,resol,n,nset,av,sig,val,dt,
bEeFitAc,bEESEF,bEENLC);
if (bPower)
power_fit(n,nset,val,t);
if (acfile) {
if (bSubAv)
for(s=0; s<nset; s++)
for(i=0; i<n; i++)
val[s][i] -= av[s];
do_autocorr(acfile,"Autocorrelation",n,nset,val,dt,
eacNormal,bAverCorr);
}
if (bRegression)
regression_analysis(n,bXYdy,t,val);
if (bLuzar)
luzar_correl(n,t,nset,val,temp,bXYdy,fit_start,smooth_tail_start);
view_all(NFILE, fnm);
thanx(stderr);
return 0;
}
int gmx_confrms(int argc, char *argv[])
{
const char *desc[] = {
"[TT]g_confrms[tt] computes the root mean square deviation (RMSD) of two",
"structures after least-squares fitting the second structure on the first one.",
"The two structures do NOT need to have the same number of atoms,",
"only the two index groups used for the fit need to be identical.",
"With [TT]-name[tt] only matching atom names from the selected groups",
"will be used for the fit and RMSD calculation. This can be useful ",
"when comparing mutants of a protein.",
"[PAR]",
"The superimposed structures are written to file. In a [TT].pdb[tt] file",
"the two structures will be written as separate models",
"(use [TT]rasmol -nmrpdb[tt]). Also in a [TT].pdb[tt] file, B-factors",
"calculated from the atomic MSD values can be written with [TT]-bfac[tt].",
};
static gmx_bool bOne = FALSE, bRmpbc = FALSE, bMW = TRUE, bName = FALSE,
bBfac = FALSE, bFit = TRUE, bLabel = FALSE;
t_pargs pa[] = {
{ "-one", FALSE, etBOOL, {&bOne}, "Only write the fitted structure to file" },
{ "-mw", FALSE, etBOOL, {&bMW}, "Mass-weighted fitting and RMSD" },
{ "-pbc", FALSE, etBOOL, {&bRmpbc}, "Try to make molecules whole again" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Do least squares superposition of the target structure to the reference" },
{ "-name", FALSE, etBOOL, {&bName},
"Only compare matching atom names" },
{ "-label", FALSE, etBOOL, {&bLabel},
"Added chain labels A for first and B for second structure"},
{ "-bfac", FALSE, etBOOL, {&bBfac},
"Output B-factors from atomic MSD values" }
};
t_filenm fnm[] = {
{ efTPS, "-f1", "conf1.gro", ffREAD },
{ efSTX, "-f2", "conf2", ffREAD },
{ efSTO, "-o", "fit.pdb", ffWRITE },
{ efNDX, "-n1", "fit1.ndx", ffOPTRD },
{ efNDX, "-n2", "fit2.ndx", ffOPTRD },
{ efNDX, "-no", "match.ndx", ffOPTWR }
};
#define NFILE asize(fnm)
/* the two structure files */
const char *conf1file, *conf2file, *matchndxfile, *outfile;
FILE *fp;
char title1[STRLEN], title2[STRLEN], *name1, *name2;
t_topology *top1, *top2;
int ePBC1, ePBC2;
t_atoms *atoms1, *atoms2;
int warn = 0;
atom_id at;
real *w_rls, mass, totmass;
rvec *x1, *v1, *x2, *v2, *fit_x;
matrix box1, box2;
output_env_t oenv;
/* counters */
int i, j, m;
/* center of mass calculation */
real tmas1, tmas2;
rvec xcm1, xcm2;
/* variables for fit */
char *groupnames1, *groupnames2;
int isize1, isize2;
atom_id *index1, *index2;
real rms, msd, minmsd, maxmsd;
real *msds;
parse_common_args(&argc, argv, PCA_BE_NICE | PCA_CAN_VIEW,
NFILE, fnm, asize(pa), pa, asize(desc), desc, 0, NULL, &oenv);
matchndxfile = opt2fn_null("-no", NFILE, fnm);
conf1file = ftp2fn(efTPS, NFILE, fnm);
conf2file = ftp2fn(efSTX, NFILE, fnm);
/* reading reference structure from first structure file */
fprintf(stderr, "\nReading first structure file\n");
snew(top1, 1);
read_tps_conf(conf1file, title1, top1, &ePBC1, &x1, &v1, box1, TRUE);
atoms1 = &(top1->atoms);
fprintf(stderr, "%s\nContaining %d atoms in %d residues\n",
title1, atoms1->nr, atoms1->nres);
if (bRmpbc)
{
rm_gropbc(atoms1, x1, box1);
}
fprintf(stderr, "Select group from first structure\n");
get_index(atoms1, opt2fn_null("-n1", NFILE, fnm),
1, &isize1, &index1, &groupnames1);
printf("\n");
if (bFit && (isize1 < 3))
{
gmx_fatal(FARGS, "Need >= 3 points to fit!\n");
}
/* reading second structure file */
fprintf(stderr, "\nReading second structure file\n");
snew(top2, 1);
read_tps_conf(conf2file, title2, top2, &ePBC2, &x2, &v2, box2, TRUE);
atoms2 = &(top2->atoms);
fprintf(stderr, "%s\nContaining %d atoms in %d residues\n",
title2, atoms2->nr, atoms2->nres);
if (bRmpbc)
{
rm_gropbc(atoms2, x2, box2);
}
fprintf(stderr, "Select group from second structure\n");
get_index(atoms2, opt2fn_null("-n2", NFILE, fnm),
1, &isize2, &index2, &groupnames2);
if (bName)
{
find_matching_names(&isize1, index1, atoms1, &isize2, index2, atoms2);
if (matchndxfile)
{
fp = ffopen(matchndxfile, "w");
fprintf(fp, "; Matching atoms between %s from %s and %s from %s\n",
groupnames1, conf1file, groupnames2, conf2file);
fprintf(fp, "[ Match_%s_%s ]\n", conf1file, groupnames1);
for (i = 0; i < isize1; i++)
{
fprintf(fp, "%4u%s", index1[i]+1, (i%15 == 14 || i == isize1-1) ? "\n" : " ");
}
fprintf(fp, "[ Match_%s_%s ]\n", conf2file, groupnames2);
for (i = 0; i < isize2; i++)
{
fprintf(fp, "%4u%s", index2[i]+1, (i%15 == 14 || i == isize2-1) ? "\n" : " ");
}
}
}
/* check isizes, must be equal */
if (isize2 != isize1)
{
gmx_fatal(FARGS, "You selected groups with differen number of atoms.\n");
}
for (i = 0; i < isize1; i++)
{
name1 = *atoms1->atomname[index1[i]];
name2 = *atoms2->atomname[index2[i]];
if (strcmp(name1, name2))
{
if (warn < 20)
{
fprintf(stderr,
"Warning: atomnames at index %d don't match: %u %s, %u %s\n",
i+1, index1[i]+1, name1, index2[i]+1, name2);
}
warn++;
}
if (!bMW)
{
atoms1->atom[index1[i]].m = 1;
atoms2->atom[index2[i]].m = 1;
}
}
if (warn)
{
fprintf(stderr, "%d atomname%s did not match\n", warn, (warn == 1) ? "" : "s");
}
if (bFit)
{
/* calculate and remove center of mass of structures */
calc_rm_cm(isize1, index1, atoms1, x1, xcm1);
calc_rm_cm(isize2, index2, atoms2, x2, xcm2);
snew(w_rls, atoms2->nr);
snew(fit_x, atoms2->nr);
for (at = 0; (at < isize1); at++)
{
w_rls[index2[at]] = atoms1->atom[index1[at]].m;
copy_rvec(x1[index1[at]], fit_x[index2[at]]);
}
/* do the least squares fit to the reference structure */
do_fit(atoms2->nr, w_rls, fit_x, x2);
sfree(fit_x);
sfree(w_rls);
w_rls = NULL;
}
else
{
clear_rvec(xcm1);
clear_rvec(xcm2);
w_rls = NULL;
}
/* calculate the rms deviation */
rms = 0;
totmass = 0;
maxmsd = -1e18;
minmsd = 1e18;
snew(msds, isize1);
for (at = 0; at < isize1; at++)
{
mass = atoms1->atom[index1[at]].m;
for (m = 0; m < DIM; m++)
{
msd = sqr(x1[index1[at]][m] - x2[index2[at]][m]);
rms += msd*mass;
msds[at] += msd;
}
maxmsd = max(maxmsd, msds[at]);
minmsd = min(minmsd, msds[at]);
totmass += mass;
}
rms = sqrt(rms/totmass);
printf("Root mean square deviation after lsq fit = %g nm\n", rms);
if (bBfac)
{
printf("Atomic MSD's range from %g to %g nm^2\n", minmsd, maxmsd);
}
if (bFit)
{
/* reset coordinates of reference and fitted structure */
for (i = 0; i < atoms1->nr; i++)
{
for (m = 0; m < DIM; m++)
{
x1[i][m] += xcm1[m];
}
}
for (i = 0; i < atoms2->nr; i++)
{
for (m = 0; m < DIM; m++)
{
x2[i][m] += xcm1[m];
}
}
}
outfile = ftp2fn(efSTO, NFILE, fnm);
switch (fn2ftp(outfile))
{
case efPDB:
case efBRK:
case efENT:
if (bBfac || bLabel)
{
srenew(atoms1->pdbinfo, atoms1->nr);
srenew(atoms1->atom, atoms1->nr); /* Why renew atom? */
/* Avoid segfaults when writing the pdb-file */
for (i = 0; i < atoms1->nr; i++)
{
atoms1->pdbinfo[i].type = eptAtom;
atoms1->pdbinfo[i].occup = 1.00;
atoms1->pdbinfo[i].bAnisotropic = FALSE;
if (bBfac)
{
atoms1->pdbinfo[i].bfac = 0;
}
if (bLabel)
{
atoms1->resinfo[atoms1->atom[i].resind].chainid = 'A';
}
}
for (i = 0; i < isize1; i++)
{
/* atoms1->pdbinfo[index1[i]].type = eptAtom; */
/* atoms1->pdbinfo[index1[i]].bAnisotropic = FALSE; */
if (bBfac)
{
atoms1->pdbinfo[index1[i]].bfac = (800*M_PI*M_PI/3.0)*msds[i];
}
/* if (bLabel) */
/* atoms1->resinfo[atoms1->atom[index1[i]].resind].chain = 'A'; */
}
srenew(atoms2->pdbinfo, atoms2->nr);
srenew(atoms2->atom, atoms2->nr); /* Why renew atom? */
for (i = 0; i < atoms2->nr; i++)
{
atoms2->pdbinfo[i].type = eptAtom;
atoms2->pdbinfo[i].occup = 1.00;
atoms2->pdbinfo[i].bAnisotropic = FALSE;
if (bBfac)
{
atoms2->pdbinfo[i].bfac = 0;
}
if (bLabel)
{
atoms2->resinfo[atoms1->atom[i].resind].chainid = 'B';
}
}
for (i = 0; i < isize2; i++)
{
/* atoms2->pdbinfo[index2[i]].type = eptAtom; */
/* atoms2->pdbinfo[index2[i]].bAnisotropic = FALSE; */
if (bBfac)
{
atoms2->pdbinfo[index2[i]].bfac = (800*M_PI*M_PI/3.0)*msds[i];
}
/* if (bLabel) */
/* atoms2->resinfo[atoms2->atom[index2[i]].resind].chain = 'B'; */
}
}
fp = ffopen(outfile, "w");
if (!bOne)
{
write_pdbfile(fp, title1, atoms1, x1, ePBC1, box1, ' ', 1, NULL, TRUE);
}
write_pdbfile(fp, title2, atoms2, x2, ePBC2, box2, ' ', bOne ? -1 : 2, NULL, TRUE);
ffclose(fp);
break;
case efGRO:
if (bBfac)
{
fprintf(stderr, "WARNING: cannot write B-factor values to gro file\n");
}
fp = ffopen(outfile, "w");
if (!bOne)
{
write_hconf_p(fp, title1, atoms1, 3, x1, v1, box1);
}
write_hconf_p(fp, title2, atoms2, 3, x2, v2, box2);
ffclose(fp);
break;
default:
if (bBfac)
{
fprintf(stderr, "WARNING: cannot write B-factor values to %s file\n",
ftp2ext(fn2ftp(outfile)));
}
if (!bOne)
{
fprintf(stderr,
"WARNING: cannot write the reference structure to %s file\n",
ftp2ext(fn2ftp(outfile)));
}
write_sto_conf(outfile, title2, atoms2, x2, v2, ePBC2, box2);
break;
}
view_all(oenv, NFILE, fnm);
thanx(stderr);
return 0;
}
int gmx_morph(int argc, char *argv[])
{
const char *desc[] = {
"[THISMODULE] does a linear interpolation of conformations in order to",
"create intermediates. Of course these are completely unphysical, but",
"that you may try to justify yourself. Output is in the form of a ",
"generic trajectory. The number of intermediates can be controlled with",
"the [TT]-ninterm[tt] flag. The first and last flag correspond to the way of",
"interpolating: 0 corresponds to input structure 1 while",
"1 corresponds to input structure 2.",
"If you specify [TT]-first[tt] < 0 or [TT]-last[tt] > 1 extrapolation will be",
"on the path from input structure x[SUB]1[sub] to x[SUB]2[sub]. In general, the coordinates",
"of the intermediate x(i) out of N total intermediates correspond to:[PAR]",
"x(i) = x[SUB]1[sub] + (first+(i/(N-1))*(last-first))*(x[SUB]2[sub]-x[SUB]1[sub])[PAR]",
"Finally the RMSD with respect to both input structures can be computed",
"if explicitly selected ([TT]-or[tt] option). In that case, an index file may be",
"read to select the group from which the RMS is computed."
};
t_filenm fnm[] = {
{ efSTX, "-f1", "conf1", ffREAD },
{ efSTX, "-f2", "conf2", ffREAD },
{ efTRX, "-o", "interm", ffWRITE },
{ efXVG, "-or", "rms-interm", ffOPTWR },
{ efNDX, "-n", "index", ffOPTRD }
};
#define NFILE asize(fnm)
static int ninterm = 11;
static real first = 0.0;
static real last = 1.0;
static gmx_bool bFit = TRUE;
t_pargs pa [] = {
{ "-ninterm", FALSE, etINT, {&ninterm},
"Number of intermediates" },
{ "-first", FALSE, etREAL, {&first},
"Corresponds to first generated structure (0 is input x[SUB]1[sub], see above)" },
{ "-last", FALSE, etREAL, {&last},
"Corresponds to last generated structure (1 is input x[SUB]2[sub], see above)" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Do a least squares fit of the second to the first structure before interpolating" }
};
const char *leg[] = { "Ref = 1\\Sst\\N conf", "Ref = 2\\Snd\\N conf" };
FILE *fp = NULL;
int i, isize, is_lsq, nat1, nat2;
t_trxstatus *status;
atom_id *index, *index_lsq, *index_all, *dummy;
t_atoms atoms;
rvec *x1, *x2, *xx, *v;
matrix box;
real rms1, rms2, fac, *mass;
char title[STRLEN], *grpname;
gmx_bool bRMS;
output_env_t oenv;
if (!parse_common_args(&argc, argv, PCA_CAN_VIEW,
NFILE, fnm, asize(pa), pa, asize(desc), desc,
0, NULL, &oenv))
{
return 0;
}
get_stx_coordnum (opt2fn("-f1", NFILE, fnm), &nat1);
get_stx_coordnum (opt2fn("-f2", NFILE, fnm), &nat2);
if (nat1 != nat2)
{
gmx_fatal(FARGS, "Number of atoms in first structure is %d, in second %d",
nat1, nat2);
}
init_t_atoms(&atoms, nat1, TRUE);
snew(x1, nat1);
snew(x2, nat1);
snew(xx, nat1);
snew(v, nat1);
read_stx_conf(opt2fn("-f1", NFILE, fnm), title, &atoms, x1, v, NULL, box);
read_stx_conf(opt2fn("-f2", NFILE, fnm), title, &atoms, x2, v, NULL, box);
snew(mass, nat1);
snew(index_all, nat1);
for (i = 0; (i < nat1); i++)
{
mass[i] = 1;
index_all[i] = i;
}
if (bFit)
{
printf("Select group for LSQ superposition:\n");
get_index(&atoms, opt2fn_null("-n", NFILE, fnm), 1, &is_lsq, &index_lsq,
&grpname);
reset_x(is_lsq, index_lsq, nat1, index_all, x1, mass);
reset_x(is_lsq, index_lsq, nat1, index_all, x2, mass);
do_fit(nat1, mass, x1, x2);
}
bRMS = opt2bSet("-or", NFILE, fnm);
if (bRMS)
{
fp = xvgropen(opt2fn("-or", NFILE, fnm), "RMSD", "Conf", "(nm)", oenv);
xvgr_legend(fp, asize(leg), leg, oenv);
printf("Select group for RMSD calculation:\n");
get_index(&atoms, opt2fn_null("-n", NFILE, fnm), 1, &isize, &index, &grpname);
printf("You selected group %s, containing %d atoms\n", grpname, isize);
rms1 = rmsdev_ind(isize, index, mass, x1, x2);
fprintf(stderr, "RMSD between input conformations is %g nm\n", rms1);
}
snew(dummy, nat1);
for (i = 0; (i < nat1); i++)
{
dummy[i] = i;
}
status = open_trx(ftp2fn(efTRX, NFILE, fnm), "w");
for (i = 0; (i < ninterm); i++)
{
fac = dointerp(nat1, x1, x2, xx, i, ninterm, first, last);
write_trx(status, nat1, dummy, &atoms, i, fac, box, xx, NULL, NULL);
if (bRMS)
{
rms1 = rmsdev_ind(isize, index, mass, x1, xx);
rms2 = rmsdev_ind(isize, index, mass, x2, xx);
fprintf(fp, "%10g %10g %10g\n", fac, rms1, rms2);
}
}
close_trx(status);
if (bRMS)
{
gmx_ffclose(fp);
do_view(oenv, opt2fn("-or", NFILE, fnm), "-nxy");
}
return 0;
}
int gmx_filter(int argc,char *argv[])
{
const char *desc[] = {
"[TT]g_filter[tt] performs frequency filtering on a trajectory.",
"The filter shape is cos([GRK]pi[grk] t/A) + 1 from -A to +A, where A is given",
"by the option [TT]-nf[tt] times the time step in the input trajectory.",
"This filter reduces fluctuations with period A by 85%, with period",
"2*A by 50% and with period 3*A by 17% for low-pass filtering.",
"Both a low-pass and high-pass filtered trajectory can be written.[PAR]",
"Option [TT]-ol[tt] writes a low-pass filtered trajectory.",
"A frame is written every [TT]-nf[tt] input frames.",
"This ratio of filter length and output interval ensures a good",
"suppression of aliasing of high-frequency motion, which is useful for",
"making smooth movies. Also averages of properties which are linear",
"in the coordinates are preserved, since all input frames are weighted",
"equally in the output.",
"When all frames are needed, use the [TT]-all[tt] option.[PAR]",
"Option [TT]-oh[tt] writes a high-pass filtered trajectory.",
"The high-pass filtered coordinates are added to the coordinates",
"from the structure file. When using high-pass filtering use [TT]-fit[tt]",
"or make sure you use a trajectory that has been fitted on",
"the coordinates in the structure file."
};
static int nf=10;
static gmx_bool bNoJump = TRUE,bFit = FALSE,bLowAll = FALSE;
t_pargs pa[] = {
{ "-nf", FALSE, etINT, {&nf},
"Sets the filter length as well as the output interval for low-pass filtering" },
{ "-all", FALSE, etBOOL, {&bLowAll},
"Write all low-pass filtered frames" },
{ "-nojump", FALSE, etBOOL, {&bNoJump},
"Remove jumps of atoms across the box" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Fit all frames to a reference structure" }
};
const char *topfile,*lowfile,*highfile;
gmx_bool bTop=FALSE;
t_topology top;
int ePBC=-1;
rvec *xtop;
matrix topbox,*box,boxf;
char title[256],*grpname;
int isize;
atom_id *index;
real *w_rls=NULL;
t_trxstatus *in;
t_trxstatus *outl,*outh;
int nffr,i,fr,nat,j,d,m;
atom_id *ind;
real flen,*filt,sum,*t;
rvec xcmtop,xcm,**x,*ptr,*xf,*xn,*xp,hbox;
output_env_t oenv;
gmx_rmpbc_t gpbc=NULL;
#define NLEG asize(leg)
t_filenm fnm[] = {
{ efTRX, "-f", NULL, ffREAD },
{ efTPS, NULL, NULL, ffOPTRD },
{ efNDX, NULL, NULL, ffOPTRD },
{ efTRO, "-ol", "lowpass", ffOPTWR },
{ efTRO, "-oh", "highpass", ffOPTWR }
};
#define NFILE asize(fnm)
parse_common_args(&argc,argv,PCA_CAN_TIME | PCA_CAN_VIEW | PCA_BE_NICE,
NFILE,fnm,asize(pa),pa,asize(desc),desc,0,NULL,&oenv);
highfile = opt2fn_null("-oh",NFILE,fnm);
if (highfile) {
topfile = ftp2fn(efTPS,NFILE,fnm);
lowfile = opt2fn_null("-ol",NFILE,fnm);
} else {
topfile = ftp2fn_null(efTPS,NFILE,fnm);
lowfile = opt2fn("-ol",NFILE,fnm);
}
if (topfile) {
bTop = read_tps_conf(ftp2fn(efTPS,NFILE,fnm),title,&top,&ePBC,
&xtop,NULL,topbox,TRUE);
if (bTop) {
gpbc = gmx_rmpbc_init(&top.idef,ePBC,top.atoms.nr,topbox);
gmx_rmpbc(gpbc,top.atoms.nr,topbox,xtop);
}
}
clear_rvec(xcmtop);
if (bFit) {
fprintf(stderr,"Select group for least squares fit\n");
get_index(&top.atoms,ftp2fn_null(efNDX,NFILE,fnm),1,&isize,&index,&grpname);
/* Set the weight */
snew(w_rls,top.atoms.nr);
for(i=0; i<isize; i++)
w_rls[index[i]] = top.atoms.atom[index[i]].m;
calc_xcm(xtop,isize,index,top.atoms.atom,xcmtop,FALSE);
for(j=0; j<top.atoms.nr; j++)
rvec_dec(xtop[j],xcmtop);
}
/* The actual filter length flen can actually be any real number */
flen = 2*nf;
/* nffr is the number of frames that we filter over */
nffr = 2*nf - 1;
snew(filt,nffr);
sum = 0;
for(i=0; i<nffr; i++) {
filt[i] = cos(2*M_PI*(i - nf + 1)/(real)flen) + 1;
sum += filt[i];
}
fprintf(stdout,"filter weights:");
for(i=0; i<nffr; i++) {
filt[i] /= sum;
fprintf(stdout," %5.3f",filt[i]);
}
fprintf(stdout,"\n");
snew(t,nffr);
snew(x,nffr);
snew(box,nffr);
nat = read_first_x(oenv,&in,opt2fn("-f",NFILE,fnm),
&(t[nffr - 1]),&(x[nffr - 1]),box[nffr - 1]);
snew(ind,nat);
for(i=0; i<nat; i++)
ind[i] = i;
/* x[nffr - 1] was already allocated by read_first_x */
for(i=0; i<nffr-1; i++)
snew(x[i],nat);
snew(xf,nat);
if (lowfile)
outl = open_trx(lowfile,"w");
else
outl = 0;
if (highfile)
outh = open_trx(highfile,"w");
else
outh = 0;
fr = 0;
do {
xn = x[nffr - 1];
if (bNoJump && fr > 0) {
xp = x[nffr - 2];
for(j=0; j<nat; j++)
for(d=0; d<DIM; d++)
hbox[d] = 0.5*box[nffr - 1][d][d];
for(i=0; i<nat; i++)
for(m=DIM-1; m>=0; m--)
if (hbox[m] > 0) {
while (xn[i][m] - xp[i][m] <= -hbox[m])
for(d=0; d<=m; d++)
xn[i][d] += box[nffr - 1][m][d];
while (xn[i][m] - xp[i][m] > hbox[m])
for(d=0; d<=m; d++)
xn[i][d] -= box[nffr - 1][m][d];
}
}
if (bTop) {
gmx_rmpbc(gpbc,nat,box[nffr - 1],xn);
}
if (bFit) {
calc_xcm(xn,isize,index,top.atoms.atom,xcm,FALSE);
for(j=0; j<nat; j++)
rvec_dec(xn[j],xcm);
do_fit(nat,w_rls,xtop,xn);
for(j=0; j<nat; j++)
rvec_inc(xn[j],xcmtop);
}
if (fr >= nffr && (outh || bLowAll || fr % nf == nf - 1)) {
/* Lowpass filtering */
for(j=0; j<nat; j++)
clear_rvec(xf[j]);
clear_mat(boxf);
for(i=0; i<nffr; i++) {
for(j=0; j<nat; j++)
for(d=0; d<DIM; d++)
xf[j][d] += filt[i]*x[i][j][d];
for(j=0; j<DIM; j++)
for(d=0; d<DIM; d++)
boxf[j][d] += filt[i]*box[i][j][d];
}
if (outl && (bLowAll || fr % nf == nf - 1))
write_trx(outl,nat,ind,topfile ? &(top.atoms) : NULL,
0,t[nf - 1],bFit ? topbox : boxf,xf,NULL,NULL);
if (outh) {
/* Highpass filtering */
for(j=0; j<nat; j++)
for(d=0; d<DIM; d++)
xf[j][d] = xtop[j][d] + x[nf - 1][j][d] - xf[j][d];
if (bFit)
for(j=0; j<nat; j++)
rvec_inc(xf[j],xcmtop);
for(j=0; j<DIM; j++)
for(d=0; d<DIM; d++)
boxf[j][d] = topbox[j][d] + box[nf - 1][j][d] - boxf[j][d];
write_trx(outh,nat,ind,topfile ? &(top.atoms) : NULL,
0,t[nf - 1],bFit ? topbox : boxf,xf,NULL,NULL);
}
}
/* Cycle all the pointer and the box by one */
ptr = x[0];
for(i=0; i<nffr-1; i++) {
t[i] = t[i+1];
x[i] = x[i+1];
copy_mat(box[i+1],box[i]);
}
x[nffr - 1] = ptr;
fr++;
} while (read_next_x(oenv,in,&(t[nffr - 1]),nat,x[nffr - 1],box[nffr - 1]));
if (bTop)
gmx_rmpbc_done(gpbc);
if (outh)
close_trx(outh);
if (outl)
close_trx(outl);
close_trx(in);
return 0;
}
static int command()
{
FILE *fp;
int i;
/* string holding name of save or load file */
char sv_file[MAX_LINE_LEN + 1];
for (i = 0; i < MAX_NUM_VAR; i++)
c_dummy_var[i][0] = NUL; /* no dummy variables */
if (is_definition(c_token))
define();
else if (almost_equals(c_token, "h$elp") || equals(c_token, "?")) {
c_token++;
do_help(1);
} else if (equals(c_token, "testtime")) {
/* given a format and a time string, exercise the time code */
char format[160], string[160];
struct tm tm;
double secs;
if (isstring(++c_token)) {
quote_str(format, c_token, 159);
if (isstring(++c_token)) {
quote_str(string, c_token++, 159);
memset(&tm, 0, sizeof(tm));
gstrptime(string, format, &tm);
secs = gtimegm(&tm);
fprintf(stderr, "internal = %f - %d/%d/%d::%d:%d:%d , wday=%d, yday=%d\n",
secs, tm.tm_mday, tm.tm_mon + 1, tm.tm_year % 100, tm.tm_hour, tm.tm_min, tm.tm_sec, tm.tm_wday, tm.tm_yday);
memset(&tm, 0, sizeof(tm));
ggmtime(&tm, secs);
gstrftime(string, 159, format, secs);
fprintf(stderr, "convert back \"%s\" - %d/%d/%d::%d:%d:%d , wday=%d, yday=%d\n",
string, tm.tm_mday, tm.tm_mon + 1, tm.tm_year % 100, tm.tm_hour, tm.tm_min, tm.tm_sec, tm.tm_wday, tm.tm_yday);
}
}
} else if (almost_equals(c_token, "test")) {
c_token++;
test_term();
} else if (almost_equals(c_token, "scr$eendump")) {
c_token++;
#ifdef _Windows
screen_dump();
#else
fputs("screendump not implemented\n", stderr);
#endif
} else if (almost_equals(c_token, "pa$use")) {
struct value a;
int sleep_time, text = 0;
char buf[MAX_LINE_LEN + 1];
c_token++;
sleep_time = (int) real(const_express(&a));
buf[0] = NUL;
if (!(END_OF_COMMAND)) {
if (!isstring(c_token))
int_error("expecting string", c_token);
else {
quote_str(buf, c_token, MAX_LINE_LEN);
++c_token;
#ifdef _Windows
if (sleep_time >= 0)
#else
# ifdef OS2
if (strcmp(term->name, "pm") != 0 || sleep_time >= 0)
# else
# ifdef MTOS
if (strcmp(term->name, "mtos") != 0 || sleep_time >= 0)
# endif /* MTOS */
# endif /* OS2 */
#endif /* _Windows */
fputs(buf, stderr);
text = 1;
}
}
if (sleep_time < 0) {
#ifdef _Windows
if (!Pause(buf))
bail_to_command_line();
#else
# ifdef OS2
if (strcmp(term->name, "pm") == 0 && sleep_time < 0) {
int rc;
if ((rc = PM_pause(buf)) == 0)
bail_to_command_line();
else if (rc == 2) {
fputs(buf, stderr);
text = 1;
(void) fgets(buf, MAX_LINE_LEN, stdin);
}
}
# else /* !OS2 */
# ifdef _Macintosh
if (strcmp(term->name, "macintosh") == 0 && sleep_time < 0)
Pause(sleep_time);
# else /* !_Macintosh */
# ifdef MTOS
if (strcmp(term->name, "mtos") == 0) {
int MTOS_pause(char *buf);
int rc;
if ((rc = MTOS_pause(buf)) == 0)
bail_to_command_line();
else if (rc == 2) {
fputs(buf, stderr);
text = 1;
(void) fgets(buf, MAX_LINE_LEN, stdin);
}
} else if (strcmp(term->name, "atari") == 0) {
char *readline(char *);
char *line = readline("");
if (line)
free(line);
} else
(void) fgets(buf, MAX_LINE_LEN, stdin);
# else /* !MTOS */
# ifdef ATARI
if (strcmp(term->name, "atari") == 0) {
char *readline(char *);
char *line = readline("");
if (line)
free(line);
} else
(void) fgets(buf, MAX_LINE_LEN, stdin);
# else /* !ATARI */
(void) fgets(buf, MAX_LINE_LEN, stdin);
/* Hold until CR hit. */
# endif /* !ATARI */
# endif /* !MTOS */
# endif /* !_Macintosh */
# endif /* !OS2 */
#endif
}
if (sleep_time > 0)
GP_SLEEP(sleep_time);
if (text != 0 && sleep_time >= 0)
fputc('\n', stderr);
screen_ok = FALSE;
} else if (almost_equals(c_token, "pr$int")) {
int need_space = 0; /* space printed between two expressions only */
screen_ok = FALSE;
do {
++c_token;
if (isstring(c_token)) {
char s[MAX_LINE_LEN];
quote_str(s, c_token, MAX_LINE_LEN);
fputs(s, stderr);
need_space = 0;
++c_token;
} else {
struct value a;
(void) const_express(&a);
if (need_space)
putc(' ', stderr);
need_space = 1;
disp_value(stderr, &a);
}
} while (!END_OF_COMMAND && equals(c_token, ","));
(void) putc('\n', stderr);
} else if (almost_equals(c_token, "fit")) {
++c_token;
do_fit();
} else if (almost_equals(c_token, "up$date")) {
char tmps[80];
char tmps2[80];
/* Have to initialise tmps2, otherwise
* update() cannot decide whether a valid
* filename was given. lh
*/
tmps2[0] = NUL;
if (!isstring(++c_token))
int_error("Parameter filename expected", c_token);
quote_str(tmps, c_token++, 80);
if (!(END_OF_COMMAND)) {
if (!isstring(c_token))
int_error("New parameter filename expected", c_token);
else
quote_str(tmps2, c_token++, 80);
}
update(tmps, tmps2);
} else if (almost_equals(c_token, "p$lot")) {
plot_token = c_token++;
SET_CURSOR_WAIT;
plotrequest();
SET_CURSOR_ARROW;
} else if (almost_equals(c_token, "sp$lot")) {
plot_token = c_token++;
SET_CURSOR_WAIT;
plot3drequest();
SET_CURSOR_ARROW;
} else if (almost_equals(c_token, "rep$lot")) {
if (replot_line[0] == NUL)
int_error("no previous plot", c_token);
c_token++;
SET_CURSOR_WAIT;
replotrequest();
SET_CURSOR_ARROW;
} else if (almost_equals(c_token, "se$t"))
set_command();
else if (almost_equals(c_token, "res$et"))
reset_command();
else if (almost_equals(c_token, "sh$ow"))
show_command();
else if (almost_equals(c_token, "cl$ear")) {
term_start_plot();
if (multiplot && term->fillbox) {
unsigned int x1 = (unsigned int) (xoffset * term->xmax);
unsigned int y1 = (unsigned int) (yoffset * term->ymax);
unsigned int width = (unsigned int) (xsize * term->xmax);
unsigned int height = (unsigned int) (ysize * term->ymax);
(*term->fillbox) (0, x1, y1, width, height);
}
term_end_plot();
screen_ok = FALSE;
c_token++;
} else if (almost_equals(c_token, "she$ll")) {
do_shell();
screen_ok = FALSE;
c_token++;
} else if (almost_equals(c_token, "sa$ve")) {
if (almost_equals(++c_token, "f$unctions")) {
if (!isstring(++c_token))
int_error("expecting filename", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
save_functions(fopen(sv_file, "w"));
}
} else if (almost_equals(c_token, "v$ariables")) {
if (!isstring(++c_token))
int_error("expecting filename", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
save_variables(fopen(sv_file, "w"));
}
} else if (almost_equals(c_token, "s$et")) {
if (!isstring(++c_token))
int_error("expecting filename", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
save_set(fopen(sv_file, "w"));
}
} else if (isstring(c_token)) {
quote_str(sv_file, c_token, MAX_LINE_LEN);
save_all(fopen(sv_file, "w"));
} else {
int_error("filename or keyword 'functions', 'variables', or 'set' expected", c_token);
}
c_token++;
} else if (almost_equals(c_token, "l$oad")) {
if (!isstring(++c_token))
int_error("expecting filename", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
/* load_file(fp=fopen(sv_file, "r"), sv_file, FALSE); OLD
* DBT 10/6/98 handle stdin as special case
* passes it on to load_file() so that it gets
* pushed on the stack and recusion will work, etc
*/
fp = strcmp(sv_file, "-") ? fopen(sv_file, "r") : stdin;
load_file(fp, sv_file, FALSE);
/* input_line[] and token[] now destroyed! */
c_token = num_tokens = 0;
}
} else if (almost_equals(c_token, "ca$ll")) {
if (!isstring(++c_token))
int_error("expecting filename", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
load_file(fopen(sv_file, "r"), sv_file, TRUE); /* Argument list follows filename */
/* input_line[] and token[] now destroyed! */
c_token = num_tokens = 0;
}
} else if (almost_equals(c_token, "if")) {
double exprval;
struct value t;
if (!equals(++c_token, "(")) /* no expression */
int_error("expecting (expression)", c_token);
exprval = real(const_express(&t));
if (exprval != 0.0) {
/* fake the condition of a ';' between commands */
int eolpos = token[num_tokens - 1].start_index + token[num_tokens - 1].length;
--c_token;
token[c_token].length = 1;
token[c_token].start_index = eolpos + 2;
input_line[eolpos + 2] = ';';
input_line[eolpos + 3] = NUL;
} else
c_token = num_tokens = 0;
} else if (almost_equals(c_token, "rer$ead")) {
fp = lf_top();
if (fp != (FILE *) NULL)
rewind(fp);
c_token++;
} else if (almost_equals(c_token, "cd")) {
if (!isstring(++c_token))
int_error("expecting directory name", c_token);
else {
quote_str(sv_file, c_token, MAX_LINE_LEN);
if (changedir(sv_file)) {
int_error("Can't change to this directory", c_token);
}
c_token++;
}
} else if (almost_equals(c_token, "pwd")) {
GP_GETCWD(sv_file, sizeof(sv_file));
fprintf(stderr, "%s\n", sv_file);
c_token++;
} else if (almost_equals(c_token, "ex$it") ||
almost_equals(c_token, "q$uit")) {
/* graphics will be tidied up in main */
return (1);
} else if (!equals(c_token, ";")) { /* null statement */
#ifdef OS2
if (_osmode == OS2_MODE) {
if (token[c_token].is_token) {
int rc;
rc = ExecuteMacro(input_line + token[c_token].start_index,
token[c_token].length);
if (rc == 0) {
c_token = num_tokens = 0;
return (0);
}
}
}
#endif
int_error("invalid command", c_token);
}
return (0);
}
void spectrum(bool bVerbose,
char *trj,char *shifts,bool bAbInitio,
char *corrfn,char *noefn,
int maxframes,bool bFour,bool bFit,int nrestart,
int npair,t_pair pair[],int nat,real chem_shifts[],
real taum,real maxdist,
real w_rls[],rvec xp[],t_idef *idef)
{
FILE *fp;
int i,j,m,ii,jj,natoms,status,nframes;
rvec *x,dx;
matrix box;
real t0,t1,t,dt;
real r2,r6,r_3,r_6,tauc;
rvec **corr;
real **Corr;
t_sij *spec;
snew(spec,npair);
fprintf(stderr,"There is no kill like overkill! Going to malloc %d bytes\n",
npair*maxframes*sizeof(corr[0][0]));
snew(corr,npair);
for(i=0; (i<npair); i++)
snew(corr[i],maxframes);
nframes = 0;
natoms = read_first_x(&status,trj,&t0,&x,box);
if (natoms > nat)
gmx_fatal(FARGS,"Not enough atoms in trajectory");
do {
if (nframes >= maxframes) {
fprintf(stderr,"\nThere are more than the %d frames you told me!",
maxframes);
break;
}
t1 = t;
if (bVerbose)
fprintf(stderr,"\rframe: %d",nframes);
rm_pbc(idef,natoms,box,x,x);
if (bFit)
do_fit(natoms,w_rls,xp,x);
for(i=0; (i<npair); i++) {
ii = pair[i].ai;
jj = pair[i].aj;
rvec_sub(x[ii],x[jj],dx);
copy_rvec(dx,corr[i][nframes]);
r2 = iprod(dx,dx);
r6 = r2*r2*r2;
r_3 = invsqrt(r6);
r_6 = r_3*r_3;
spec[i].rij_3 += r_3;
spec[i].rij_6 += r_6;
for(m=0; (m<5); m++) {
spec[i].Ylm[m] = c_add(spec[i].Ylm[m],
calc_ylm(m-2,dx,r2,r_3,r_6));
}
}
nframes++;
} while (read_next_x(status,&t,natoms,x,box));
close_trj(status);
if (bVerbose)
fprintf(stderr,"\n");
fp=ffopen("ylm.out","w");
calc_aver(fp,nframes,npair,pair,spec,maxdist);
fclose(fp);
/* Select out the pairs that have to be correlated */
snew(Corr,npair);
for(i=j=0; (i<npair); i++) {
if (spec[i].bNOE) {
Corr[j] = &(corr[i][0][0]);
j++;
}
}
fprintf(stderr,"There are %d NOEs in your simulation\n",j);
if (nframes > 1)
dt = (t1-t0)/(nframes-1);
else
dt = 1;
do_autocorr(corrfn,"Correlation Function for Interproton Vectors",
nframes,j,Corr,dt,eacP2,nrestart,FALSE,FALSE,bFour,TRUE);
calc_tauc(bVerbose,npair,pair,dt,nframes/2,spec,(real **)corr);
plot_spectrum(noefn,npair,pair,spec,taum);
}
int gmx_cluster(int argc,char *argv[])
{
static char *desc[] = {
"g_cluster can cluster structures with several different methods.",
"Distances between structures can be determined from a trajectory",
"or read from an XPM matrix file with the [TT]-dm[tt] option.",
"RMS deviation after fitting or RMS deviation of atom-pair distances",
"can be used to define the distance between structures.[PAR]",
"single linkage: add a structure to a cluster when its distance to any",
"element of the cluster is less than [TT]cutoff[tt].[PAR]",
"Jarvis Patrick: add a structure to a cluster when this structure",
"and a structure in the cluster have each other as neighbors and",
"they have a least [TT]P[tt] neighbors in common. The neighbors",
"of a structure are the M closest structures or all structures within",
"[TT]cutoff[tt].[PAR]",
"Monte Carlo: reorder the RMSD matrix using Monte Carlo.[PAR]",
"diagonalization: diagonalize the RMSD matrix.[PAR]"
"gromos: use algorithm as described in Daura [IT]et al.[it]",
"([IT]Angew. Chem. Int. Ed.[it] [BB]1999[bb], [IT]38[it], pp 236-240).",
"Count number of neighbors using cut-off, take structure with",
"largest number of neighbors with all its neighbors as cluster",
"and eleminate it from the pool of clusters. Repeat for remaining",
"structures in pool.[PAR]",
"When the clustering algorithm assigns each structure to exactly one",
"cluster (single linkage, Jarvis Patrick and gromos) and a trajectory",
"file is supplied, the structure with",
"the smallest average distance to the others or the average structure",
"or all structures for each cluster will be written to a trajectory",
"file. When writing all structures, separate numbered files are made",
"for each cluster.[PAR]"
"Two output files are always written:[BR]",
"[TT]-o[tt] writes the RMSD values in the upper left half of the matrix",
"and a graphical depiction of the clusters in the lower right half",
"When [TT]-minstruct[tt] = 1 the graphical depiction is black",
"when two structures are in the same cluster.",
"When [TT]-minstruct[tt] > 1 different colors will be used for each",
"cluster.[BR]",
"[TT]-g[tt] writes information on the options used and a detailed list",
"of all clusters and their members.[PAR]",
"Additionally, a number of optional output files can be written:[BR]",
"[TT]-dist[tt] writes the RMSD distribution.[BR]",
"[TT]-ev[tt] writes the eigenvectors of the RMSD matrix",
"diagonalization.[BR]",
"[TT]-sz[tt] writes the cluster sizes.[BR]",
"[TT]-tr[tt] writes a matrix of the number transitions between",
"cluster pairs.[BR]",
"[TT]-ntr[tt] writes the total number of transitions to or from",
"each cluster.[BR]",
"[TT]-clid[tt] writes the cluster number as a function of time.[BR]",
"[TT]-cl[tt] writes average (with option [TT]-av[tt]) or central",
"structure of each cluster or writes numbered files with cluster members",
"for a selected set of clusters (with option [TT]-wcl[tt], depends on",
"[TT]-nst[tt] and [TT]-rmsmin[tt]).[BR]",
};
FILE *fp,*log;
int i,i1,i2,j,nf,nrms;
matrix box;
rvec *xtps,*usextps,*x1,**xx=NULL;
char *fn,*trx_out_fn;
t_clusters clust;
t_mat *rms;
real *eigval;
t_topology top;
int ePBC;
t_atoms useatoms;
t_matrix *readmat;
real *tmp;
int isize=0,ifsize=0,iosize=0;
atom_id *index=NULL, *fitidx, *outidx;
char *grpname;
real rmsd,**d1,**d2,*time,time_invfac,*mass=NULL;
char buf[STRLEN],buf1[80],title[STRLEN];
bool bAnalyze,bUseRmsdCut,bJP_RMSD=FALSE,bReadMat,bReadTraj;
int method,ncluster=0;
static char *methodname[] = {
NULL, "linkage", "jarvis-patrick","monte-carlo",
"diagonalization", "gromos", NULL
};
enum { m_null, m_linkage, m_jarvis_patrick,
m_monte_carlo, m_diagonalize, m_gromos, m_nr };
/* Set colors for plotting: white = zero RMS, black = maximum */
static t_rgb rlo_top = { 1.0, 1.0, 1.0 };
static t_rgb rhi_top = { 0.0, 0.0, 0.0 };
static t_rgb rlo_bot = { 1.0, 1.0, 1.0 };
static t_rgb rhi_bot = { 0.0, 0.0, 1.0 };
static int nlevels=40,skip=1;
static real scalemax=-1.0,rmsdcut=0.1,rmsmin=0.0;
static bool bRMSdist=FALSE,bBinary=FALSE,bAverage=FALSE,bFit=TRUE;
static int niter=10000,seed=1993,write_ncl=0,write_nst=1,minstruct=1;
static real kT=1e-3;
static int M=10,P=3;
t_pargs pa[] = {
{ "-dista", FALSE, etBOOL, {&bRMSdist},
"Use RMSD of distances instead of RMS deviation" },
{ "-nlevels",FALSE,etINT, {&nlevels},
"Discretize RMSD matrix in # levels" },
{ "-cutoff",FALSE, etREAL, {&rmsdcut},
"RMSD cut-off (nm) for two structures to be neighbor" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Use least squares fitting before RMSD calculation" },
{ "-max", FALSE, etREAL, {&scalemax},
"Maximum level in RMSD matrix" },
{ "-skip", FALSE, etINT, {&skip},
"Only analyze every nr-th frame" },
{ "-av", FALSE, etBOOL, {&bAverage},
"Write average iso middle structure for each cluster" },
{ "-wcl", FALSE, etINT, {&write_ncl},
"Write all structures for first # clusters to numbered files" },
{ "-nst", FALSE, etINT, {&write_nst},
"Only write all structures if more than # per cluster" },
{ "-rmsmin",FALSE, etREAL, {&rmsmin},
"minimum rms difference with rest of cluster for writing structures" },
{ "-method",FALSE, etENUM, {methodname},
"Method for cluster determination" },
{ "-minstruct", FALSE, etINT, {&minstruct},
"Minimum number of structures in cluster for coloring in the xpm file" },
{ "-binary",FALSE, etBOOL, {&bBinary},
"Treat the RMSD matrix as consisting of 0 and 1, where the cut-off "
"is given by -cutoff" },
{ "-M", FALSE, etINT, {&M},
"Number of nearest neighbors considered for Jarvis-Patrick algorithm, "
"0 is use cutoff" },
{ "-P", FALSE, etINT, {&P},
"Number of identical nearest neighbors required to form a cluster" },
{ "-seed", FALSE, etINT, {&seed},
"Random number seed for Monte Carlo clustering algorithm" },
{ "-niter", FALSE, etINT, {&niter},
"Number of iterations for MC" },
{ "-kT", FALSE, etREAL, {&kT},
"Boltzmann weighting factor for Monte Carlo optimization "
"(zero turns off uphill steps)" }
};
t_filenm fnm[] = {
{ efTRX, "-f", NULL, ffOPTRD },
{ efTPS, "-s", NULL, ffOPTRD },
{ efNDX, NULL, NULL, ffOPTRD },
{ efXPM, "-dm", "rmsd", ffOPTRD },
{ efXPM, "-o", "rmsd-clust", ffWRITE },
{ efLOG, "-g", "cluster", ffWRITE },
{ efXVG, "-dist", "rmsd-dist", ffOPTWR },
{ efXVG, "-ev", "rmsd-eig", ffOPTWR },
{ efXVG, "-sz", "clust-size", ffOPTWR},
{ efXPM, "-tr", "clust-trans",ffOPTWR},
{ efXVG, "-ntr", "clust-trans",ffOPTWR},
{ efXVG, "-clid", "clust-id.xvg",ffOPTWR},
{ efTRX, "-cl", "clusters.pdb", ffOPTWR }
};
#define NFILE asize(fnm)
CopyRight(stderr,argv[0]);
parse_common_args(&argc,argv,PCA_CAN_VIEW | PCA_CAN_TIME | PCA_TIME_UNIT | PCA_BE_NICE,
NFILE,fnm,asize(pa),pa,asize(desc),desc,0,NULL);
/* parse options */
bReadMat = opt2bSet("-dm",NFILE,fnm);
bReadTraj = opt2bSet("-f",NFILE,fnm) || !bReadMat;
if ( opt2parg_bSet("-av",asize(pa),pa) ||
opt2parg_bSet("-wcl",asize(pa),pa) ||
opt2parg_bSet("-nst",asize(pa),pa) ||
opt2parg_bSet("-rmsmin",asize(pa),pa) ||
opt2bSet("-cl",NFILE,fnm) )
trx_out_fn = opt2fn("-cl",NFILE,fnm);
else
trx_out_fn = NULL;
if (bReadMat && time_factor()!=1) {
fprintf(stderr,
"\nWarning: assuming the time unit in %s is %s\n",
opt2fn("-dm",NFILE,fnm),time_unit());
}
if (trx_out_fn && !bReadTraj)
fprintf(stderr,"\nWarning: "
"cannot write cluster structures without reading trajectory\n"
" ignoring option -cl %s\n", trx_out_fn);
method=1;
while ( method < m_nr && strcasecmp(methodname[0], methodname[method])!=0 )
method++;
if (method == m_nr)
gmx_fatal(FARGS,"Invalid method");
bAnalyze = (method == m_linkage || method == m_jarvis_patrick ||
method == m_gromos );
/* Open log file */
log = ftp2FILE(efLOG,NFILE,fnm,"w");
fprintf(stderr,"Using %s method for clustering\n",methodname[0]);
fprintf(log,"Using %s method for clustering\n",methodname[0]);
/* check input and write parameters to log file */
bUseRmsdCut = FALSE;
if (method == m_jarvis_patrick) {
bJP_RMSD = (M == 0) || opt2parg_bSet("-cutoff",asize(pa),pa);
if ((M<0) || (M == 1))
gmx_fatal(FARGS,"M (%d) must be 0 or larger than 1",M);
if (M < 2) {
sprintf(buf1,"Will use P=%d and RMSD cutoff (%g)",P,rmsdcut);
bUseRmsdCut = TRUE;
} else {
if (P >= M)
gmx_fatal(FARGS,"Number of neighbors required (P) must be less than M");
if (bJP_RMSD) {
sprintf(buf1,"Will use P=%d, M=%d and RMSD cutoff (%g)",P,M,rmsdcut);
bUseRmsdCut = TRUE;
} else
sprintf(buf1,"Will use P=%d, M=%d",P,M);
}
ffprintf1(stderr,log,buf,"%s for determining the neighbors\n\n",buf1);
} else /* method != m_jarvis */
bUseRmsdCut = ( bBinary || method == m_linkage || method == m_gromos );
if (bUseRmsdCut && method != m_jarvis_patrick)
fprintf(log,"Using RMSD cutoff %g nm\n",rmsdcut);
if ( method==m_monte_carlo )
fprintf(log,"Using %d iterations\n",niter);
if (skip < 1)
gmx_fatal(FARGS,"skip (%d) should be >= 1",skip);
/* get input */
if (bReadTraj) {
/* don't read mass-database as masses (and top) are not used */
read_tps_conf(ftp2fn(efTPS,NFILE,fnm),buf,&top,&ePBC,&xtps,NULL,box,
bAnalyze);
fprintf(stderr,"\nSelect group for least squares fit%s:\n",
bReadMat?"":" and RMSD calculation");
get_index(&(top.atoms),ftp2fn_null(efNDX,NFILE,fnm),
1,&ifsize,&fitidx,&grpname);
if (trx_out_fn) {
fprintf(stderr,"\nSelect group for output:\n");
get_index(&(top.atoms),ftp2fn_null(efNDX,NFILE,fnm),
1,&iosize,&outidx,&grpname);
/* merge and convert both index groups: */
/* first copy outidx to index. let outidx refer to elements in index */
snew(index,iosize);
isize = iosize;
for(i=0; i<iosize; i++) {
index[i]=outidx[i];
outidx[i]=i;
}
/* now lookup elements from fitidx in index, add them if necessary
and also let fitidx refer to elements in index */
for(i=0; i<ifsize; i++) {
j=0;
while (j<isize && index[j]!=fitidx[i])
j++;
if (j>=isize) {
/* slow this way, but doesn't matter much */
isize++;
srenew(index,isize);
}
index[j]=fitidx[i];
fitidx[i]=j;
}
} else { /* !trx_out_fn */
isize = ifsize;
snew(index, isize);
for(i=0; i<ifsize; i++) {
index[i]=fitidx[i];
fitidx[i]=i;
}
}
}
/* Initiate arrays */
snew(d1,isize);
snew(d2,isize);
for(i=0; (i<isize); i++) {
snew(d1[i],isize);
snew(d2[i],isize);
}
if (bReadTraj) {
/* Loop over first coordinate file */
fn = opt2fn("-f",NFILE,fnm);
xx = read_whole_trj(fn,isize,index,skip,&nf,&time);
convert_times(nf, time);
if (!bRMSdist || bAnalyze) {
/* Center all frames on zero */
snew(mass,isize);
for(i=0; i<ifsize; i++)
mass[fitidx[i]] = top.atoms.atom[index[fitidx[i]]].m;
if (bFit)
for(i=0; i<nf; i++)
reset_x(ifsize,fitidx,isize,NULL,xx[i],mass);
}
}
if (bReadMat) {
fprintf(stderr,"Reading rms distance matrix ");
read_xpm_matrix(opt2fn("-dm",NFILE,fnm),&readmat);
fprintf(stderr,"\n");
if (readmat[0].nx != readmat[0].ny)
gmx_fatal(FARGS,"Matrix (%dx%d) is not square",
readmat[0].nx,readmat[0].ny);
if (bReadTraj && bAnalyze && (readmat[0].nx != nf))
gmx_fatal(FARGS,"Matrix size (%dx%d) does not match the number of "
"frames (%d)",readmat[0].nx,readmat[0].ny,nf);
nf = readmat[0].nx;
sfree(time);
time = readmat[0].axis_x;
time_invfac = time_invfactor();
for(i=0; i<nf; i++)
time[i] *= time_invfac;
rms = init_mat(readmat[0].nx,method == m_diagonalize);
convert_mat(&(readmat[0]),rms);
nlevels = readmat[0].nmap;
} else { /* !bReadMat */
rms = init_mat(nf,method == m_diagonalize);
nrms = (nf*(nf-1))/2;
if (!bRMSdist) {
fprintf(stderr,"Computing %dx%d RMS deviation matrix\n",nf,nf);
snew(x1,isize);
for(i1=0; (i1<nf); i1++) {
for(i2=i1+1; (i2<nf); i2++) {
for(i=0; i<isize; i++)
copy_rvec(xx[i1][i],x1[i]);
if (bFit)
do_fit(isize,mass,xx[i2],x1);
rmsd = rmsdev(isize,mass,xx[i2],x1);
set_mat_entry(rms,i1,i2,rmsd);
}
nrms -= (nf-i1-1);
fprintf(stderr,"\r# RMSD calculations left: %d ",nrms);
}
} else { /* bRMSdist */
fprintf(stderr,"Computing %dx%d RMS distance deviation matrix\n",nf,nf);
for(i1=0; (i1<nf); i1++) {
calc_dist(isize,xx[i1],d1);
for(i2=i1+1; (i2<nf); i2++) {
calc_dist(isize,xx[i2],d2);
set_mat_entry(rms,i1,i2,rms_dist(isize,d1,d2));
}
nrms -= (nf-i1-1);
fprintf(stderr,"\r# RMSD calculations left: %d ",nrms);
}
}
fprintf(stderr,"\n\n");
}
ffprintf2(stderr,log,buf,"The RMSD ranges from %g to %g nm\n",
rms->minrms,rms->maxrms);
ffprintf1(stderr,log,buf,"Average RMSD is %g\n",2*rms->sumrms/(nf*(nf-1)));
ffprintf1(stderr,log,buf,"Number of structures for matrix %d\n",nf);
ffprintf1(stderr,log,buf,"Energy of the matrix is %g nm\n",mat_energy(rms));
if (bUseRmsdCut && (rmsdcut < rms->minrms || rmsdcut > rms->maxrms) )
fprintf(stderr,"WARNING: rmsd cutoff %g is outside range of rmsd values "
"%g to %g\n",rmsdcut,rms->minrms,rms->maxrms);
if (bAnalyze && (rmsmin < rms->minrms) )
fprintf(stderr,"WARNING: rmsd minimum %g is below lowest rmsd value %g\n",
rmsmin,rms->minrms);
if (bAnalyze && (rmsmin > rmsdcut) )
fprintf(stderr,"WARNING: rmsd minimum %g is above rmsd cutoff %g\n",
rmsmin,rmsdcut);
/* Plot the rmsd distribution */
rmsd_distribution(opt2fn("-dist",NFILE,fnm),rms);
if (bBinary) {
for(i1=0; (i1 < nf); i1++)
for(i2=0; (i2 < nf); i2++)
if (rms->mat[i1][i2] < rmsdcut)
rms->mat[i1][i2] = 0;
else
rms->mat[i1][i2] = 1;
}
snew(clust.cl,nf);
switch (method) {
case m_linkage:
/* Now sort the matrix and write it out again */
gather(rms,rmsdcut,&clust);
break;
case m_diagonalize:
/* Do a diagonalization */
snew(eigval,nf);
snew(tmp,nf*nf);
memcpy(tmp,rms->mat[0],nf*nf*sizeof(real));
eigensolver(tmp,nf,0,nf,eigval,rms->mat[0]);
sfree(tmp);
fp = xvgropen(opt2fn("-ev",NFILE,fnm),"RMSD matrix Eigenvalues",
"Eigenvector index","Eigenvalues (nm\\S2\\N)");
for(i=0; (i<nf); i++)
fprintf(fp,"%10d %10g\n",i,eigval[i]);
ffclose(fp);
break;
case m_monte_carlo:
mc_optimize(log,rms,niter,&seed,kT);
swap_mat(rms);
reset_index(rms);
break;
case m_jarvis_patrick:
jarvis_patrick(rms->nn,rms->mat,M,P,bJP_RMSD ? rmsdcut : -1,&clust);
break;
case m_gromos:
gromos(rms->nn,rms->mat,rmsdcut,&clust);
break;
default:
gmx_fatal(FARGS,"DEATH HORROR unknown method \"%s\"",methodname[0]);
}
if (method == m_monte_carlo || method == m_diagonalize)
fprintf(stderr,"Energy of the matrix after clustering is %g nm\n",
mat_energy(rms));
if (bAnalyze) {
if (minstruct > 1) {
ncluster = plot_clusters(nf,rms->mat,&clust,nlevels,minstruct);
} else {
mark_clusters(nf,rms->mat,rms->maxrms,&clust);
}
init_t_atoms(&useatoms,isize,FALSE);
snew(usextps, isize);
useatoms.resname=top.atoms.resname;
for(i=0; i<isize; i++) {
useatoms.atomname[i]=top.atoms.atomname[index[i]];
useatoms.atom[i].resnr=top.atoms.atom[index[i]].resnr;
useatoms.nres=max(useatoms.nres,useatoms.atom[i].resnr+1);
copy_rvec(xtps[index[i]],usextps[i]);
}
useatoms.nr=isize;
analyze_clusters(nf,&clust,rms->mat,isize,&useatoms,usextps,mass,xx,time,
ifsize,fitidx,iosize,outidx,
bReadTraj?trx_out_fn:NULL,
opt2fn_null("-sz",NFILE,fnm),
opt2fn_null("-tr",NFILE,fnm),
opt2fn_null("-ntr",NFILE,fnm),
opt2fn_null("-clid",NFILE,fnm),
bAverage, write_ncl, write_nst, rmsmin, bFit, log,
rlo_bot,rhi_bot);
}
ffclose(log);
if (bBinary && !bAnalyze)
/* Make the clustering visible */
for(i2=0; (i2 < nf); i2++)
for(i1=i2+1; (i1 < nf); i1++)
if (rms->mat[i1][i2])
rms->mat[i1][i2] = rms->maxrms;
fp = opt2FILE("-o",NFILE,fnm,"w");
fprintf(stderr,"Writing rms distance/clustering matrix ");
if (bReadMat) {
write_xpm(fp,0,readmat[0].title,readmat[0].legend,readmat[0].label_x,
readmat[0].label_y,nf,nf,readmat[0].axis_x,readmat[0].axis_y,
rms->mat,0.0,rms->maxrms,rlo_top,rhi_top,&nlevels);
}
else {
sprintf(buf,"Time (%s)",time_unit());
sprintf(title,"RMS%sDeviation / Cluster Index",
bRMSdist ? " Distance " : " ");
if (minstruct > 1) {
write_xpm_split(fp,0,title,"RMSD (nm)",buf,buf,
nf,nf,time,time,rms->mat,0.0,rms->maxrms,&nlevels,
rlo_top,rhi_top,0.0,(real) ncluster,
&ncluster,TRUE,rlo_bot,rhi_bot);
} else {
write_xpm(fp,0,title,"RMSD (nm)",buf,buf,
nf,nf,time,time,rms->mat,0.0,rms->maxrms,
rlo_top,rhi_top,&nlevels);
}
}
fprintf(stderr,"\n");
ffclose(fp);
/* now show what we've done */
do_view(opt2fn("-o",NFILE,fnm),"-nxy");
do_view(opt2fn_null("-sz",NFILE,fnm),"-nxy");
if (method == m_diagonalize)
do_view(opt2fn_null("-ev",NFILE,fnm),"-nxy");
do_view(opt2fn("-dist",NFILE,fnm),"-nxy");
if (bAnalyze) {
do_view(opt2fn_null("-tr",NFILE,fnm),"-nxy");
do_view(opt2fn_null("-ntr",NFILE,fnm),"-nxy");
do_view(opt2fn_null("-clid",NFILE,fnm),"-nxy");
}
/* Thank the user for her patience */
thanx(stderr);
return 0;
}
int gmx_rmsf(int argc, char *argv[])
{
const char *desc[] = {
"[THISMODULE] computes the root mean square fluctuation (RMSF, i.e. standard ",
"deviation) of atomic positions in the trajectory (supplied with [TT]-f[tt])",
"after (optionally) fitting to a reference frame (supplied with [TT]-s[tt]).[PAR]",
"With option [TT]-oq[tt] the RMSF values are converted to B-factor",
"values, which are written to a [REF].pdb[ref] file with the coordinates, of the",
"structure file, or of a [REF].pdb[ref] file when [TT]-q[tt] is specified.",
"Option [TT]-ox[tt] writes the B-factors to a file with the average",
"coordinates.[PAR]",
"With the option [TT]-od[tt] the root mean square deviation with",
"respect to the reference structure is calculated.[PAR]",
"With the option [TT]-aniso[tt], [THISMODULE] will compute anisotropic",
"temperature factors and then it will also output average coordinates",
"and a [REF].pdb[ref] file with ANISOU records (corresonding to the [TT]-oq[tt]",
"or [TT]-ox[tt] option). Please note that the U values",
"are orientation-dependent, so before comparison with experimental data",
"you should verify that you fit to the experimental coordinates.[PAR]",
"When a [REF].pdb[ref] input file is passed to the program and the [TT]-aniso[tt]",
"flag is set",
"a correlation plot of the Uij will be created, if any anisotropic",
"temperature factors are present in the [REF].pdb[ref] file.[PAR]",
"With option [TT]-dir[tt] the average MSF (3x3) matrix is diagonalized.",
"This shows the directions in which the atoms fluctuate the most and",
"the least."
};
static gmx_bool bRes = FALSE, bAniso = FALSE, bFit = TRUE;
t_pargs pargs[] = {
{ "-res", FALSE, etBOOL, {&bRes},
"Calculate averages for each residue" },
{ "-aniso", FALSE, etBOOL, {&bAniso},
"Compute anisotropic termperature factors" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Do a least squares superposition before computing RMSF. Without this you must make sure that the reference structure and the trajectory match." }
};
int natom;
int i, m, teller = 0;
real t, *w_rls;
t_topology top;
int ePBC;
t_atoms *pdbatoms, *refatoms;
matrix box, pdbbox;
rvec *x, *pdbx, *xref;
t_trxstatus *status;
const char *label;
FILE *fp; /* the graphics file */
const char *devfn, *dirfn;
int resind;
gmx_bool bReadPDB;
int *index;
int isize;
char *grpnames;
real bfac, pdb_bfac, *Uaver;
double **U, *xav;
int aid;
rvec *rmsd_x = nullptr;
double *rmsf, invcount, totmass;
int d;
real count = 0;
rvec xcm;
gmx_rmpbc_t gpbc = nullptr;
gmx_output_env_t *oenv;
const char *leg[2] = { "MD", "X-Ray" };
t_filenm fnm[] = {
{ efTRX, "-f", nullptr, ffREAD },
{ efTPS, nullptr, nullptr, ffREAD },
{ efNDX, nullptr, nullptr, ffOPTRD },
{ efPDB, "-q", nullptr, ffOPTRD },
{ efPDB, "-oq", "bfac", ffOPTWR },
{ efPDB, "-ox", "xaver", ffOPTWR },
{ efXVG, "-o", "rmsf", ffWRITE },
{ efXVG, "-od", "rmsdev", ffOPTWR },
{ efXVG, "-oc", "correl", ffOPTWR },
{ efLOG, "-dir", "rmsf", ffOPTWR }
};
#define NFILE asize(fnm)
if (!parse_common_args(&argc, argv, PCA_CAN_TIME | PCA_CAN_VIEW,
NFILE, fnm, asize(pargs), pargs, asize(desc), desc, 0, nullptr,
&oenv))
{
return 0;
}
bReadPDB = ftp2bSet(efPDB, NFILE, fnm);
devfn = opt2fn_null("-od", NFILE, fnm);
dirfn = opt2fn_null("-dir", NFILE, fnm);
read_tps_conf(ftp2fn(efTPS, NFILE, fnm), &top, &ePBC, &xref, nullptr, box, TRUE);
const char *title = *top.name;
snew(w_rls, top.atoms.nr);
fprintf(stderr, "Select group(s) for root mean square calculation\n");
get_index(&top.atoms, ftp2fn_null(efNDX, NFILE, fnm), 1, &isize, &index, &grpnames);
/* Set the weight */
for (i = 0; i < isize; i++)
{
w_rls[index[i]] = top.atoms.atom[index[i]].m;
}
/* Malloc the rmsf arrays */
snew(xav, isize*DIM);
snew(U, isize);
for (i = 0; i < isize; i++)
{
snew(U[i], DIM*DIM);
}
snew(rmsf, isize);
if (devfn)
{
snew(rmsd_x, isize);
}
if (bReadPDB)
{
t_topology *top_pdb;
snew(top_pdb, 1);
/* Read coordinates twice */
read_tps_conf(opt2fn("-q", NFILE, fnm), top_pdb, nullptr, nullptr, nullptr, pdbbox, FALSE);
snew(pdbatoms, 1);
*pdbatoms = top_pdb->atoms;
read_tps_conf(opt2fn("-q", NFILE, fnm), top_pdb, nullptr, &pdbx, nullptr, pdbbox, FALSE);
/* TODO Should this assert that top_pdb->atoms.nr == top.atoms.nr?
* See discussion at https://gerrit.gromacs.org/#/c/6430/1 */
title = *top_pdb->name;
snew(refatoms, 1);
*refatoms = top_pdb->atoms;
sfree(top_pdb);
}
else
{
pdbatoms = &top.atoms;
refatoms = &top.atoms;
pdbx = xref;
snew(pdbatoms->pdbinfo, pdbatoms->nr);
pdbatoms->havePdbInfo = TRUE;
copy_mat(box, pdbbox);
}
if (bFit)
{
sub_xcm(xref, isize, index, top.atoms.atom, xcm, FALSE);
}
natom = read_first_x(oenv, &status, ftp2fn(efTRX, NFILE, fnm), &t, &x, box);
if (bFit)
{
gpbc = gmx_rmpbc_init(&top.idef, ePBC, natom);
}
/* Now read the trj again to compute fluctuations */
teller = 0;
do
{
if (bFit)
{
/* Remove periodic boundary */
gmx_rmpbc(gpbc, natom, box, x);
/* Set center of mass to zero */
sub_xcm(x, isize, index, top.atoms.atom, xcm, FALSE);
/* Fit to reference structure */
do_fit(natom, w_rls, xref, x);
}
/* Calculate Anisotropic U Tensor */
for (i = 0; i < isize; i++)
{
aid = index[i];
for (d = 0; d < DIM; d++)
{
xav[i*DIM + d] += x[aid][d];
for (m = 0; m < DIM; m++)
{
U[i][d*DIM + m] += x[aid][d]*x[aid][m];
}
}
}
if (devfn)
{
/* Calculate RMS Deviation */
for (i = 0; (i < isize); i++)
{
aid = index[i];
for (d = 0; (d < DIM); d++)
{
rmsd_x[i][d] += gmx::square(x[aid][d]-xref[aid][d]);
}
}
}
count += 1.0;
teller++;
}
while (read_next_x(oenv, status, &t, x, box));
close_trx(status);
if (bFit)
{
gmx_rmpbc_done(gpbc);
}
invcount = 1.0/count;
snew(Uaver, DIM*DIM);
totmass = 0;
for (i = 0; i < isize; i++)
{
for (d = 0; d < DIM; d++)
{
xav[i*DIM + d] *= invcount;
}
for (d = 0; d < DIM; d++)
{
for (m = 0; m < DIM; m++)
{
U[i][d*DIM + m] = U[i][d*DIM + m]*invcount
- xav[i*DIM + d]*xav[i*DIM + m];
Uaver[3*d+m] += top.atoms.atom[index[i]].m*U[i][d*DIM + m];
}
}
totmass += top.atoms.atom[index[i]].m;
}
for (d = 0; d < DIM*DIM; d++)
{
Uaver[d] /= totmass;
}
if (bRes)
{
for (d = 0; d < DIM*DIM; d++)
{
average_residues(nullptr, U, d, isize, index, w_rls, &top.atoms);
}
}
if (bAniso)
{
for (i = 0; i < isize; i++)
{
aid = index[i];
pdbatoms->pdbinfo[aid].bAnisotropic = TRUE;
pdbatoms->pdbinfo[aid].uij[U11] = static_cast<int>(1e6*U[i][XX*DIM + XX]);
pdbatoms->pdbinfo[aid].uij[U22] = static_cast<int>(1e6*U[i][YY*DIM + YY]);
pdbatoms->pdbinfo[aid].uij[U33] = static_cast<int>(1e6*U[i][ZZ*DIM + ZZ]);
pdbatoms->pdbinfo[aid].uij[U12] = static_cast<int>(1e6*U[i][XX*DIM + YY]);
pdbatoms->pdbinfo[aid].uij[U13] = static_cast<int>(1e6*U[i][XX*DIM + ZZ]);
pdbatoms->pdbinfo[aid].uij[U23] = static_cast<int>(1e6*U[i][YY*DIM + ZZ]);
}
}
if (bRes)
{
label = "Residue";
}
else
{
label = "Atom";
}
for (i = 0; i < isize; i++)
{
rmsf[i] = U[i][XX*DIM + XX] + U[i][YY*DIM + YY] + U[i][ZZ*DIM + ZZ];
}
if (dirfn)
{
fprintf(stdout, "\n");
print_dir(stdout, Uaver);
fp = gmx_ffopen(dirfn, "w");
print_dir(fp, Uaver);
gmx_ffclose(fp);
}
for (i = 0; i < isize; i++)
{
sfree(U[i]);
}
sfree(U);
/* Write RMSF output */
if (bReadPDB)
{
bfac = 8.0*M_PI*M_PI/3.0*100;
fp = xvgropen(ftp2fn(efXVG, NFILE, fnm), "B-Factors",
label, "(A\\b\\S\\So\\N\\S2\\N)", oenv);
xvgr_legend(fp, 2, leg, oenv);
for (i = 0; (i < isize); i++)
{
if (!bRes || i+1 == isize ||
top.atoms.atom[index[i]].resind != top.atoms.atom[index[i+1]].resind)
{
resind = top.atoms.atom[index[i]].resind;
pdb_bfac = find_pdb_bfac(pdbatoms, &top.atoms.resinfo[resind],
*(top.atoms.atomname[index[i]]));
fprintf(fp, "%5d %10.5f %10.5f\n",
bRes ? top.atoms.resinfo[top.atoms.atom[index[i]].resind].nr : index[i]+1, rmsf[i]*bfac,
pdb_bfac);
}
}
xvgrclose(fp);
}
else
{
fp = xvgropen(ftp2fn(efXVG, NFILE, fnm), "RMS fluctuation", label, "(nm)", oenv);
for (i = 0; i < isize; i++)
{
if (!bRes || i+1 == isize ||
top.atoms.atom[index[i]].resind != top.atoms.atom[index[i+1]].resind)
{
fprintf(fp, "%5d %8.4f\n",
bRes ? top.atoms.resinfo[top.atoms.atom[index[i]].resind].nr : index[i]+1, std::sqrt(rmsf[i]));
}
}
xvgrclose(fp);
}
for (i = 0; i < isize; i++)
{
pdbatoms->pdbinfo[index[i]].bfac = 800*M_PI*M_PI/3.0*rmsf[i];
}
if (devfn)
{
for (i = 0; i < isize; i++)
{
rmsf[i] = (rmsd_x[i][XX]+rmsd_x[i][YY]+rmsd_x[i][ZZ])/count;
}
if (bRes)
{
average_residues(rmsf, nullptr, 0, isize, index, w_rls, &top.atoms);
}
/* Write RMSD output */
fp = xvgropen(devfn, "RMS Deviation", label, "(nm)", oenv);
for (i = 0; i < isize; i++)
{
if (!bRes || i+1 == isize ||
top.atoms.atom[index[i]].resind != top.atoms.atom[index[i+1]].resind)
{
fprintf(fp, "%5d %8.4f\n",
bRes ? top.atoms.resinfo[top.atoms.atom[index[i]].resind].nr : index[i]+1, std::sqrt(rmsf[i]));
}
}
xvgrclose(fp);
}
if (opt2bSet("-oq", NFILE, fnm))
{
/* Write a .pdb file with B-factors and optionally anisou records */
for (i = 0; i < isize; i++)
{
rvec_inc(pdbx[index[i]], xcm);
}
write_sto_conf_indexed(opt2fn("-oq", NFILE, fnm), title, pdbatoms, pdbx,
nullptr, ePBC, pdbbox, isize, index);
}
if (opt2bSet("-ox", NFILE, fnm))
{
rvec *bFactorX;
snew(bFactorX, top.atoms.nr);
for (i = 0; i < isize; i++)
{
for (d = 0; d < DIM; d++)
{
bFactorX[index[i]][d] = xcm[d] + xav[i*DIM + d];
}
}
/* Write a .pdb file with B-factors and optionally anisou records */
write_sto_conf_indexed(opt2fn("-ox", NFILE, fnm), title, pdbatoms, bFactorX, nullptr,
ePBC, pdbbox, isize, index);
sfree(bFactorX);
}
if (bAniso)
{
correlate_aniso(opt2fn("-oc", NFILE, fnm), refatoms, pdbatoms, oenv);
do_view(oenv, opt2fn("-oc", NFILE, fnm), "-nxy");
}
do_view(oenv, opt2fn("-o", NFILE, fnm), "-nxy");
if (devfn)
{
do_view(oenv, opt2fn("-od", NFILE, fnm), "-nxy");
}
return 0;
}
int gmx_rmsf(int argc,char *argv[])
{
static char *desc[] = {
"g_rmsf computes the root mean square fluctuation (RMSF, i.e. standard ",
"deviation) of atomic positions ",
"after (optionally) fitting to a reference frame.[PAR]",
"With option [TT]-oq[tt] the RMSF values are converted to B-factor",
"values, which are written to a pdb file with the coordinates, of the",
"structure file, or of a pdb file when [TT]-q[tt] is specified.",
"Option [TT]-ox[tt] writes the B-factors to a file with the average",
"coordinates.[PAR]",
"With the option [TT]-od[tt] the root mean square deviation with",
"respect to the reference structure is calculated.[PAR]",
"With the option [TT]aniso[tt] g_rmsf will compute anisotropic",
"temperature factors and then it will also output average coordinates",
"and a pdb file with ANISOU records (corresonding to the [TT]-oq[tt]",
"or [TT]-ox[tt] option). Please note that the U values",
"are orientation dependent, so before comparison with experimental data",
"you should verify that you fit to the experimental coordinates.[PAR]",
"When a pdb input file is passed to the program and the [TT]-aniso[tt]",
"flag is set",
"a correlation plot of the Uij will be created, if any anisotropic",
"temperature factors are present in the pdb file.[PAR]",
"With option [TT]-dir[tt] the average MSF (3x3) matrix is diagonalized.",
"This shows the directions in which the atoms fluctuate the most and",
"the least."
};
static bool bRes=FALSE,bAniso=FALSE,bdevX=FALSE,bFit=TRUE;
t_pargs pargs[] = {
{ "-res", FALSE, etBOOL, {&bRes},
"Calculate averages for each residue" },
{ "-aniso",FALSE, etBOOL, {&bAniso},
"Compute anisotropic termperature factors" },
{ "-fit", FALSE, etBOOL, {&bFit},
"Do a least squares superposition before computing RMSF. Without this you must make sure that the reference structure and the trajectory match." }
};
int step,nre,natom,natoms,i,g,m,teller=0;
real t,lambda,*w_rls,*w_rms;
t_tpxheader header;
t_inputrec ir;
t_topology top;
int ePBC;
t_atoms *pdbatoms,*refatoms;
bool bCont;
matrix box,pdbbox;
rvec *x,*pdbx,*xref;
int status,npdbatoms,res0;
char buf[256],*label;
char title[STRLEN];
FILE *fp; /* the graphics file */
char *devfn,*dirfn;
int resnr;
bool bReadPDB;
atom_id *index;
int isize;
char *grpnames;
real bfac,pdb_bfac,*Uaver;
double **U,*xav;
atom_id aid;
rvec *rmsd_x=NULL;
real *rmsf,invcount,totmass;
int d;
real count=0;
rvec xcm;
char *leg[2] = { "MD", "X-Ray" };
t_filenm fnm[] = {
{ efTRX, "-f", NULL, ffREAD },
{ efTPS, NULL, NULL, ffREAD },
{ efNDX, NULL, NULL, ffOPTRD },
{ efPDB, "-q", NULL, ffOPTRD },
{ efPDB, "-oq", "bfac", ffOPTWR },
{ efPDB, "-ox", "xaver", ffOPTWR },
{ efXVG, "-o", "rmsf", ffWRITE },
{ efXVG, "-od", "rmsdev", ffOPTWR },
{ efXVG, "-oc", "correl", ffOPTWR },
{ efLOG, "-dir", "rmsf", ffOPTWR }
};
#define NFILE asize(fnm)
CopyRight(stderr,argv[0]);
parse_common_args(&argc,argv,PCA_CAN_TIME | PCA_CAN_VIEW | PCA_BE_NICE ,
NFILE,fnm,asize(pargs),pargs,asize(desc),desc,0,NULL);
bReadPDB = ftp2bSet(efPDB,NFILE,fnm);
devfn = opt2fn_null("-od",NFILE,fnm);
dirfn = opt2fn_null("-dir",NFILE,fnm);
read_tps_conf(ftp2fn(efTPS,NFILE,fnm),title,&top,&ePBC,&xref,NULL,box,TRUE);
snew(w_rls,top.atoms.nr);
fprintf(stderr,"Select group(s) for root mean square calculation\n");
get_index(&top.atoms,ftp2fn_null(efNDX,NFILE,fnm),1,&isize,&index,&grpnames);
/* Set the weight */
for(i=0; i<isize; i++)
w_rls[index[i]]=top.atoms.atom[index[i]].m;
/* Malloc the rmsf arrays */
snew(xav,isize*DIM);
snew(U,isize);
for(i=0; i<isize; i++)
snew(U[i],DIM*DIM);
snew(rmsf,isize);
if (devfn)
snew(rmsd_x, isize);
if (bReadPDB) {
get_stx_coordnum(opt2fn("-q",NFILE,fnm),&npdbatoms);
snew(pdbatoms,1);
snew(refatoms,1);
init_t_atoms(pdbatoms,npdbatoms,TRUE);
init_t_atoms(refatoms,npdbatoms,TRUE);
snew(pdbx,npdbatoms);
/* Read coordinates twice */
read_stx_conf(opt2fn("-q",NFILE,fnm),title,pdbatoms,pdbx,NULL,NULL,pdbbox);
read_stx_conf(opt2fn("-q",NFILE,fnm),title,refatoms,pdbx,NULL,NULL,pdbbox);
} else {
pdbatoms = &top.atoms;
refatoms = &top.atoms;
pdbx = xref;
npdbatoms = pdbatoms->nr;
snew(pdbatoms->pdbinfo,npdbatoms);
copy_mat(box,pdbbox);
}
if (bFit)
sub_xcm(xref,isize,index,top.atoms.atom,xcm,FALSE);
natom = read_first_x(&status,ftp2fn(efTRX,NFILE,fnm),&t,&x,box);
/* Now read the trj again to compute fluctuations */
teller = 0;
do {
if (bFit) {
/* Remove periodic boundary */
rm_pbc(&(top.idef),ePBC,natom,box,x,x);
/* Set center of mass to zero */
sub_xcm(x,isize,index,top.atoms.atom,xcm,FALSE);
/* Fit to reference structure */
do_fit(natom,w_rls,xref,x);
}
/* Calculate Anisotropic U Tensor */
for(i=0; i<isize; i++) {
aid = index[i];
for(d=0; d<DIM; d++) {
xav[i*DIM + d] += x[aid][d];
for(m=0; m<DIM; m++)
U[i][d*DIM + m] += x[aid][d]*x[aid][m];
}
}
if (devfn) {
/* Calculate RMS Deviation */
for(i=0;(i<isize);i++) {
aid = index[i];
for(d=0;(d<DIM);d++) {
rmsd_x[i][d] += sqr(x[aid][d]-xref[aid][d]);
}
}
}
count += 1.0;
teller++;
} while(read_next_x(status,&t,natom,x,box));
close_trj(status);
invcount = 1.0/count;
snew(Uaver,DIM*DIM);
totmass = 0;
for(i=0; i<isize; i++) {
for(d=0; d<DIM; d++)
xav[i*DIM + d] *= invcount;
for(d=0; d<DIM; d++)
for(m=0; m<DIM; m++) {
U[i][d*DIM + m] = U[i][d*DIM + m]*invcount
- xav[i*DIM + d]*xav[i*DIM + m];
Uaver[3*d+m] += top.atoms.atom[index[i]].m*U[i][d*DIM + m];
}
totmass += top.atoms.atom[index[i]].m;
}
for(d=0; d<DIM*DIM; d++)
Uaver[d] /= totmass;
if (bAniso) {
for(i=0; i<isize; i++) {
aid = index[i];
pdbatoms->pdbinfo[aid].bAnisotropic = TRUE;
pdbatoms->pdbinfo[aid].uij[U11] = 1e6*U[i][XX*DIM + XX];
pdbatoms->pdbinfo[aid].uij[U22] = 1e6*U[i][YY*DIM + YY];
pdbatoms->pdbinfo[aid].uij[U33] = 1e6*U[i][ZZ*DIM + ZZ];
pdbatoms->pdbinfo[aid].uij[U12] = 1e6*U[i][XX*DIM + YY];
pdbatoms->pdbinfo[aid].uij[U13] = 1e6*U[i][XX*DIM + ZZ];
pdbatoms->pdbinfo[aid].uij[U23] = 1e6*U[i][YY*DIM + ZZ];
}
}
if (bRes) {
average_residues(rmsf,isize,index,w_rls,&top.atoms);
label = "Residue";
} else
label = "Atom";
for(i=0; i<isize; i++)
rmsf[i] = U[i][XX*DIM + XX] + U[i][YY*DIM + YY] + U[i][ZZ*DIM + ZZ];
if (dirfn) {
fprintf(stdout,"\n");
print_dir(stdout,Uaver);
fp = ffopen(dirfn,"w");
print_dir(fp,Uaver);
fclose(fp);
}
for(i=0; i<isize; i++)
sfree(U[i]);
sfree(U);
/* Write RMSF output */
if (bReadPDB) {
bfac = 8.0*M_PI*M_PI/3.0*100;
fp = xvgropen(ftp2fn(efXVG,NFILE,fnm),"B-Factors",
label,"(A\\b\\S\\So\\N\\S2\\N)");
xvgr_legend(fp,2,leg);
for(i=0;(i<isize);i++) {
if (!bRes || i+1==isize ||
top.atoms.atom[index[i]].resnr!=top.atoms.atom[index[i+1]].resnr) {
resnr = top.atoms.atom[index[i]].resnr;
pdb_bfac = find_pdb_bfac(pdbatoms,*(top.atoms.resname[resnr]),resnr,
*(top.atoms.atomname[index[i]]));
fprintf(fp,"%5d %10.5f %10.5f\n",
bRes ? top.atoms.atom[index[i]].resnr+1 : i+1,rmsf[i]*bfac,
pdb_bfac);
}
}
fclose(fp);
} else {
fp = xvgropen(ftp2fn(efXVG,NFILE,fnm),"RMS fluctuation",label,"(nm)");
for(i=0; i<isize; i++)
if (!bRes || i+1==isize ||
top.atoms.atom[index[i]].resnr!=top.atoms.atom[index[i+1]].resnr)
fprintf(fp,"%5d %8.4f\n",
bRes ? top.atoms.atom[index[i]].resnr+1 : i+1,sqrt(rmsf[i]));
fclose(fp);
}
for(i=0; i<isize; i++)
pdbatoms->pdbinfo[index[i]].bfac = 800*M_PI*M_PI/3.0*rmsf[i];
if (devfn) {
for(i=0; i<isize; i++)
rmsf[i] = (rmsd_x[i][XX]+rmsd_x[i][YY]+rmsd_x[i][ZZ])/count;
if (bRes)
average_residues(rmsf,isize,index,w_rls,&top.atoms);
/* Write RMSD output */
fp = xvgropen(devfn,"RMS Deviation",label,"(nm)");
for(i=0; i<isize; i++)
if (!bRes || i+1==isize ||
top.atoms.atom[index[i]].resnr!=top.atoms.atom[index[i+1]].resnr)
fprintf(fp,"%5d %8.4f\n",
bRes ? top.atoms.atom[index[i]].resnr+1 : i+1,sqrt(rmsf[i]));
fclose(fp);
}
if (opt2bSet("-oq",NFILE,fnm)) {
/* Write a pdb file with B-factors and optionally anisou records */
for(i=0; i<isize; i++)
rvec_inc(xref[index[i]],xcm);
write_sto_conf_indexed(opt2fn("-oq",NFILE,fnm),title,pdbatoms,pdbx,
NULL,ePBC,pdbbox,isize,index);
}
if (opt2bSet("-ox",NFILE,fnm)) {
/* Misuse xref as a temporary array */
for(i=0; i<isize; i++)
for(d=0; d<DIM; d++)
xref[index[i]][d] = xcm[d] + xav[i*DIM + d];
/* Write a pdb file with B-factors and optionally anisou records */
write_sto_conf_indexed(opt2fn("-ox",NFILE,fnm),title,pdbatoms,xref,NULL,
ePBC,pdbbox,isize,index);
}
if (bAniso) {
correlate_aniso(opt2fn("-oc",NFILE,fnm),refatoms,pdbatoms);
do_view(opt2fn("-oc",NFILE,fnm),"-nxy");
}
do_view(opt2fn("-o",NFILE,fnm),"-nxy");
if (devfn)
do_view(opt2fn("-od",NFILE,fnm),"-nxy");
thanx(stderr);
return 0;
}
