/*********************************************************
* *
* Main Function *
* ------------- *
* *
*********************************************************
| Takes arguments and launches the gof simulations. |
*-------------------------------------------------------*/
int main(int argc, char *argv[]) // Array of char=arguments line
{
//--- Declarations & Call Function ---//
int i=0, j=0, k=0, count=0, howmany=0, segsites=0, okim=0, oksim=0, numsim=0, totsim=0, nokl=0, nokl0=0;
int **statseg=NULL, **nbvariant=NULL, oks[3], okstot[3];
FILE *pf=NULL, *fopen(const char*, const char*); // Pointer on File for outputs (pf) and IM input file
double tajd();
char **list=NULL; // Haplotype list
void updatemainparams(struct params*);
int gensam(struct params*, char**, int**, int*);
int **imatrix(int, int);
//// From rand1.c ////
/* Celine changed 03/18/2010 */
void seedit(char*, FILE*, struct params *);/*/////*/
char **cmatrix(int, int);
//// From params.c ////
void changeparams(struct params*);
void changeparamslocus(struct params*, int);
struct params getpars(int, char*[], int*);
//--- Structure declaration---//
struct params param;
//--- Get arguments ---//
param=getpars(argc, argv, &howmany); // Get input by user for parameters
pf=stdout; // Output
/* Celine changed 03/18/2010 */
if( !param.commandlineseedflag ) seedit("s", pf, ¶m);// WRITE seeds in summary output file
/*/////*/
/* Uncommented for Celine's use */
/* for(i=0;i<argc;i++) // Information on simulation
fprintf(pf, "%s ", argv[i]);
*/////\
//---------- Initialisation & Memory allocation ------------------//
nbvariant=imatrix(param.cp.npop+1, maxsites); // array of nb of frequency spectrum
typeseg=(int*)malloc((unsigned)(maxsites*sizeof(int))); // type of sites
statseg=imatrix(param.cp.npop+3, howmany); // Records locus specific S1, S2, Ss, Sf,
changeparams(¶m); // Change estimates parameters from priors
updatemainparams(¶m); // Update parameters for the coalescent
oksim=totsim=okstot[0]=okstot[1]=okstot[1]=okstot[2]=0; // simulation check, total #of sim, check on statistics for all simulations
for(numsim=0; numsim<param.cp.nsim;numsim++) // Loop along number of simulations for this set of parameters
{
//--- Initialization and reset of quality checks ---//
count=nokl=nokl0=okim=oks[0]=oks[1]=oks[1]=oks[2]=0; // number of loci, # loci with ok genealogies, no set sites, #statistics ok
for(i=0;i<11;i++) // Sim specific Stats
param.cp.sSiFst[i]=0;
//--- Loop along the loci in the simulation ---//
while((howmany-count++))
{
if(okim==0) // Case All loci ok in the sample
{
for(i=0;i<11;i++)
{
param.cp.lSiFst[i]=0.0;
if(i<9)
param.lp[count-1].tpH[i]=0;
}
changeparamslocus(¶m, count-1); // Get locus specific parameters
list=cmatrix(param.cp.nsam, maxsites+1); // Allocate list of haplotypes
segsites=gensam(¶m, list, nbvariant, param.lp[count-1].S);// Generate a new gene ARG
statseg[0][count-1]=segsites; // Total number of seg sites in sample
for(i=1;i<3+param.cp.npop;i++)
statseg[i][count-1]=0;
if((segsites>0)) // Case segsite>0: get stats
{
/* fprintf(pf, "segsites:%d\npositions:\n",segsites); */
/* for(i=0;i<param.cp.nsam;i++) fprintf(pf, "%s\n", list[i]); */
/* fprintf(pf, "\n"); */
if(segsites<=param.cp.nsites) // Case segsite < lenght of locus
{
for(k=0;k<param.cp.nsam;k++)
{
for(i=k+1;i<param.cp.nsam;i++)
{
if((k<param.lp[count-1].ni[1])&&(i<param.lp[count-1].ni[1])) // pop1
{
param.lp[count-1].tpH[0]++; // # chromosomes
for(j=0;j<segsites;j++)
{
if(list[k][j]!=list[i][j])
param.lp[count-1].tpH[1]++; // # seg sites
}
}
else if((k>=param.lp[count-1].ni[1])&&(i>=param.lp[count-1].ni[1])) // pop2
{
param.lp[count-1].tpH[2]++; // # chromosomes
for(j=0;j<segsites;j++)
{
if(list[k][j]!=list[i][j])
param.lp[count-1].tpH[3]++; // # seg sites
}
}
else // total sample
{
param.lp[count-1].tpH[4]++;; // Totsal sample size
for(j=0;j<segsites;j++)
{
if(list[k][j]!=list[i][j])
param.lp[count-1].tpH[5]++; // total S
}
}
}// Loop on chromosome
}// Loop along all sampled sequence for the locus
for(i=0;i<segsites;i++) // Calulate S statistics for the locus
{
if(typeseg[i]<0) statseg[3][count-1]++; // shared
else if(typeseg[i]<param.cp.npop+1) statseg[typeseg[i]][count-1]++; // population specific
else statseg[4][count-1]++; // fixed
}
for(i=1;i<5;i++) //--- Record S1 S2 Ss Sf forthe locus ---//
param.cp.lSiFst[i-1]+=statseg[i][count-1];
for(i=0;i<param.cp.nsam;i++) // Free memory for this locus
free(list[i]);
free(list);
}// End case segsite<lenght of locus
else // Case segsites>lenght of locus
{
okim=1; // Sample have a wrong locus
oksim=1; // stop this simulation
break;
}
}// End locus polymorphic
else // Locus without seg sites
{
okim=1; // Sample have a wrong locus (S=0)
oksim=2; // 0 for all stats
}
}// End Sample good until now
if(okim==0) // All loci good until now
{
nokl++; // +1 good locus
nokl0++; // +1 polymorphic locus
for(i=0;i<7;i++)
{
if(i<4)
param.cp.sSiFst[i]+=param.cp.lSiFst[i]; // sum of Sk
param.lp[count-1].H[i]=param.lp[count-1].tpH[i]; // locus specific stats
}
param.cp.lSiFst[5]=param.lp[count-1].H[1]/=param.lp[count-1].H[0]; // Hw1
param.cp.lSiFst[6]=param.lp[count-1].H[3]/=param.lp[count-1].H[2]; // Hw2
param.lp[count-1].H[5]/=param.lp[count-1].H[4]; // Hb
param.cp.lSiFst[4]=param.lp[count-1].H[6]=1-((param.lp[count-1].H[1]+param.lp[count-1].H[3])/2)/ param.lp[count-1].H[5];// locis specific Fst
if((param.lp[count-1].S[0]>0)&&(param.lp[count-1].ni[1]>2)) // Locus popymorphic in pop1
{
param.cp.lSiFst[7]=tajd(param.lp[count-1].ni[1], param.lp[count-1].S[0], param.lp[count-1].H[1]);
param.cp.sSiFst[7]+=param.cp.lSiFst[7];
oks[0]++; // +1 good stat for pop1
}
if((param.lp[count-1].S[1]>0)&&(param.lp[count-1].ni[2]>2)) // Locus popymorphic in pop2
{
param.cp.lSiFst[8]=tajd(param.lp[count-1].ni[2], param.lp[count-1].S[1], param.lp[count-1].H[3]);
param.cp.sSiFst[8]+=param.cp.lSiFst[8];
oks[1]++; // +1 good stat for pop2
}
if(statseg[1][count-1]>0) // Locus popymorphic private in pop1
param.lp[count-1].H[7]=param.cp.lSiFst[9]+=(double) param.lp[count-1].S[2]/(statseg[1][count-1]*param.lp[count-1].ni[1]*2); // p(1)
if(statseg[2][count-1]>0) // Locus popymorphic private in pop2
param.lp[count-1].H[7]=param.cp.lSiFst[9]+=(double) param.lp[count-1].S[3]/(statseg[2][count-1]*param.lp[count-1].ni[2]*2); // p(1)
param.cp.sSiFst[9]+=param.cp.lSiFst[9]; // sum p1
if(statseg[3][count-1]>0) // Locus popymorphic private in pop2
{
param.lp[count-1].H[8]=param.cp.lSiFst[10]=(double) param.lp[count-1].S[4]/(statseg[3][count-1]*(param.lp[count-1].ni[2]+param.lp[count-1].ni[1])); // p(2)
param.cp.sSiFst[10]+=param.cp.lSiFst[10]; // sum p2
oks[2]++;
}
param.cp.sSiFst[4]+=param.lp[count-1].H[6]; // sum Fst
param.cp.sSiFst[5]+=param.lp[count-1].H[1]; // sum Hw1
param.cp.sSiFst[6]+=param.lp[count-1].H[3]; // sum Hw2
}
else if(oksim==2) // Case no seg site for that locus
{
oksim=0; // reset checks
okim=0;
nokl++; // 1+ locus to count in mean (all 0 values)
for(i=0;i<9;i++)
param.lp[count-1].H[i]=param.lp[count-1].tpH[i]; // locus specific stats
}
}// End loop on loci
if(nokl==howmany) // All sample good
{
totsim++; // 1+ good simulation
for(i=0;i<4;i++)
param.cp.SiFst[i]+=param.cp.sSiFst[i]; // sum of S stats along simulations
for(i=4;i<11;i++)
{
if(((i<7)||(i>=9))&&(nokl0>0))
param.cp.SiFst[i]+=(double)param.cp.sSiFst[i]/nokl0; // mean of other stats along simulations
if((i>=7)&&(i<9)&&(oks[i-7]>0))
{
param.cp.SiFst[i]+=(double)param.cp.sSiFst[i]/oks[i-7];
okstot[i-7]++;
}
}
}
}// End loop on simulations
if(oksim==0) // All simulations worked
{
/* Uncommented for Celine's use */
/* for(i=0;i<11;i++) */////
for(i=0;i<9;i++)
{
if((i<7)||(i>=9))
fprintf(pf, "%lg\t", (double) param.cp.SiFst[i]/(totsim)); // write mean of sum of S stats, Fst and Hws over simulations
if((i>=7)&&(i<9))
{
if(oks[i-7]>0)
fprintf(pf, "%lg\t", (double) param.cp.SiFst[i]/(okstot[i-7])); // write mean Tds if S>0 in pops
else fprintf(pf, "NA\t" );
}
}
fprintf(pf, "\n");
}
else // Case one locus with too much seg sites
{
for(i=0;i<9;i++)
fprintf(pf, "NA\t" );
fprintf(pf, "\n");
}
/* Celine changed 03/18/2010 */
seedit("end", pf, ¶m); // in randx.c, flag[0]!="s" so create/rewrite seed in seedmimar
/*/////*/
fclose(pf);
free(typeseg);
for(i=0;i<param.cp.npop+3;i++)
{
if(i<param.cp.npop+1)
free(nbvariant[i]);
free(statseg[i]);
}
free(nbvariant);
free(statseg);
///////// FREE PARAM ///////
for(i=0;i<param.cp.npop;i++)
free(param.cp.mig_mat[i]);
free(param.cp.mig_mat);
free(param.cp.config);
/* Celine changed 11/27/2009 */
for(i=9;i>=0;i--)
if(param.cp.listevent[i]!=NULL && param.cp.listevent[i]->nextde!=NULL)
free(param.cp.listevent[i]->nextde);
if(param.cp.listevent!=NULL)
free(param.cp.listevent);
/*/////*/
free(param.cp.deventlist);
free(param.cp.size);
free(param.cp.alphag);
for(i=0;i<3;i++)
free(param.cp.uniform[i]);
free(param.cp.uniform);
free(param.cp.oldest);
free(param.cp.newest);
free(param.cp.newparam);
/* Celine changed 11/27/2009 */
for(i=0;i<howmany;i++)
free(param.lp[i].name);
/*/////*/
free(param.lp);
/* Celine changed 03/18/2010 */
free( param.tableseeds); /*/////*/
exit(0);
}// End main function
int
main(int argc, char *argv[])
{
int i, k, howmany, segsites ;
char **list, **cmatrix(), **tbsparamstrs ;
FILE *pf, *fopen() ;
double probss, tmrca, ttot ;
void seedit( const char * ) ;
void getpars( int argc, char *argv[], int *howmany ) ;
int gensam( char **list, double *probss, double *ptmrca, double *pttot ) ;
ntbs = 0 ; /* these next few lines are for reading in parameters from a file (for each sample) */
tbsparamstrs = (char **)malloc( argc*sizeof(char *) ) ;
for( i=0; i<argc; i++) printf("%s ",argv[i]);
for( i =0; i<argc; i++) tbsparamstrs[i] = (char *)malloc(30*sizeof(char) ) ;
for( i = 1; i<argc ; i++)
if( strcmp( argv[i],"tbs") == 0 ) argv[i] = tbsparamstrs[ ntbs++] ;
count=0;
if( ntbs > 0 ) for( k=0; k<ntbs; k++) scanf(" %s", tbsparamstrs[k] );
getpars( argc, argv, &howmany) ; /* results are stored in global variable, pars */
if( !pars.commandlineseedflag ) seedit( "s");
pf = stdout ;
if( pars.mp.segsitesin == 0 ) {
list = cmatrix(pars.cp.nsam,maxsites+1);
posit = (double *)malloc( (unsigned)( maxsites*sizeof( double)) ) ;
}
else {
list = cmatrix(pars.cp.nsam, pars.mp.segsitesin+1 ) ;
posit = (double *)malloc( (unsigned)( pars.mp.segsitesin*sizeof( double)) ) ;
if( pars.mp.theta > 0.0 ) {
segfac = 1.0 ;
for( i= pars.mp.segsitesin; i > 1; i--) segfac *= i ;
}
}
while( howmany-count++ ) {
if( (ntbs > 0) && (count >1 ) ) {
for( k=0; k<ntbs; k++) {
if( scanf(" %s", tbsparamstrs[k]) == EOF ) {
if( !pars.commandlineseedflag ) seedit( "end" );
exit(0);
}
}
getpars( argc, argv, &howmany) ;
}
fprintf(pf,"\n//");
if( ntbs >0 ) {
for(k=0; k< ntbs; k++) printf("\t%s", tbsparamstrs[k] ) ;
}
printf("\n");
segsites = gensam( list, &probss, &tmrca, &ttot ) ;
if( pars.mp.timeflag ) fprintf(pf,"time:\t%lf\t%lf\n",tmrca, ttot ) ;
if( (segsites > 0 ) || ( pars.mp.theta > 0.0 ) ) {
if( (pars.mp.segsitesin > 0 ) && ( pars.mp.theta > 0.0 ))
fprintf(pf,"prob: %g\n", probss ) ;
fprintf(pf,"segsites: %d\n",segsites);
if( segsites > 0 ) fprintf(pf,"positions: ");
for( i=0; i<segsites; i++)
fprintf(pf,"%6.*lf ", pars.output_precision,posit[i] );
fprintf(pf,"\n");
if( segsites > 0 )
for(i=0; i<pars.cp.nsam; i++) {
fprintf(pf,"%s\n", list[i] );
}
}
}
if( !pars.commandlineseedflag ) seedit( "end" );
}
