/* @funcstatic seqwords_keysearch ********************************************
**
** Search swissprot with terms structure and writes a hitlist structure
**
** @param [r] inf [AjPFile] File pointer to swissprot database
** @param [r] terms [AjPTerms] Terms object pointer
** @param [w] hits [EmbPHitlist*] Hitlist object pointer
**
** @return [AjBool] True on success
** @@
******************************************************************************/
static AjBool seqwords_keysearch(AjPFile inf,
AjPTerms terms,
EmbPHitlist *hits)
{
AjPStr line =NULL; /* Line of text. */
AjPStr id =NULL; /* Line of text. */
AjPStr temp =NULL;
ajint s =0; /* Temp. start of hit value. */
ajint e =0; /* Temp. end of hit value. */
AjPInt start =NULL; /* Array of start of hit(s). */
AjPInt end =NULL; /* Array of end of hit(s). */
ajint nhits =0; /* Number of hits. */
ajint x =0;
AjBool foundkw =ajFalse;
AjBool foundft =ajFalse;
/* Check for valid args. */
if(!inf)
return ajFalse;
/* Allocate strings and arrays. */
line = ajStrNew();
id = ajStrNew();
start = ajIntNew();
end = ajIntNew();
/* Start of main loop. */
while((ajReadlineTrim(inf,&line)))
{
/* Parse the AC line. */
if(ajStrPrefixC(line,"AC"))
{
/* Copy accesion number and remove the ';' from the end. */
ajFmtScanS(line, "%*s %S", &id);
ajStrExchangeCC(&id, ";", "\0");
/* Reset flags & no. hits. */
foundkw=ajFalse;
foundft=ajFalse;
nhits=0;
}
/* Search the description and keyword lines with search terms. */
else if((ajStrPrefixC(line,"DE") || (ajStrPrefixC(line,"KW"))))
{
/*
** Search terms have a leading and trailing space to prevent
** them being found as substrings within other words. To
** catch cases where a DE or KW line begins with a search
** term, we must add a leading and trailing space to line.
** We must first remove punctation from the line to be parsed.
*/
ajStrExchangeSetCC(&line, ".,;:", " ");
ajStrAppendK(&line, ' ');
ajStrInsertC(&line, 0, " ");
for (x = 0; x < terms->N; x++)
/* Search term is found. */
if((ajStrFindCaseS(line, terms->Keywords[x])!=-1))
{
foundkw=ajTrue;
break;
}
}
/* Search the feature table line with search terms. */
else if((ajStrPrefixC(line,"FT DOMAIN")))
{
/*
** Search terms have a leading and trailing space to prevent
** them being found as substrings within other words. To
** catch cases where a FT line ends with a search
** term, we must add a trailing space to line
** We must first remove punctation from the line to be parsed.
*/
ajStrExchangeSetCC(&line, ".,;:", " ");
ajStrAppendK(&line, ' ');
for (x = 0; x < terms->N; x++)
if((ajStrFindCaseS(line, terms->Keywords[x])!=-1))
{
/* Search term is found. */
foundft = ajTrue;
nhits++;
/* Assign start and end of hit. */
ajFmtScanS(line, "%*s %*s %d %d", &s, &e);
ajIntPut(&start, nhits-1, s);
ajIntPut(&end, nhits-1, e);
break;
}
}
/* Parse the sequence. */
else if((ajStrPrefixC(line,"SQ") && ((foundkw == ajTrue) ||
(foundft == ajTrue))))
{
/* Allocate memory for temp. sequence. */
temp = ajStrNew();
/* Read the sequence into hitlist structure. */
while((ajReadlineTrim(inf,&line)) && !ajStrPrefixC(line,"//"))
/* Read sequence line into temp. */
ajStrAppendC(&temp,ajStrGetPtr(line)+3);
/* Clean up temp. sequence. */
ajStrRemoveWhite(&temp);
/*Priority is given to domain (rather than full length) sequence.*/
if(foundft)
{
for(x=0;x<nhits;x++)
{
/* Increment counter of hits for subsequent hits*/
(*hits)->N++;
/* Reallocate memory for array of hits in hitlist
structure. */
AJCRESIZE((*hits)->hits, (*hits)->N);
(*hits)->hits[(*hits)->N-1]=embHitNew();
ajStrAssignC(&(*hits)->hits[(*hits)->N-1]->Model,
"KEYWORD");
/* Assign start and end of hit. */
(*hits)->hits[(*hits)->N-1]->Start = ajIntGet(start, x);
(*hits)->hits[(*hits)->N-1]->End = ajIntGet(end, x);
/* Extract sequence within specified range */
ajStrAssignSubS(&(*hits)->hits[(*hits)->N - 1]->Seq, temp,
(*hits)->hits[(*hits)->N - 1]->Start - 1,
(*hits)->hits[(*hits)->N - 1]->End - 1);
/* Put id into structure */
ajStrAssignRef(&(*hits)->hits[(*hits)->N - 1]->Acc, id);
}
}
else
{
/* Increment counter of hits */
(*hits)->N++;
/* Reallocate memory for array of hits in hitlist structure */
AJCRESIZE((*hits)->hits, (*hits)->N);
(*hits)->hits[(*hits)->N-1]=embHitNew();
ajStrAssignC(&(*hits)->hits[(*hits)->N-1]->Model, "KEYWORD");
/* Extract whole sequence */
ajStrAssignRef(&(*hits)->hits[(*hits)->N - 1]->Seq, temp);
(*hits)->hits[(*hits)->N - 1]->Start = 1;
(*hits)->hits[(*hits)->N - 1]->End =
ajStrGetLen((*hits)->hits[(*hits)->N - 1]->Seq);
/* Put id into structure */
ajStrAssignRef(&(*hits)->hits[(*hits)->N - 1]->Acc, id);
}
/* Free temp. sequence */
ajStrDel(&temp);
}
}
/* Clean up */
ajStrDel(&line);
ajStrDel(&id);
ajIntDel(&start);
ajIntDel(&end);
return ajTrue;
}
int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPFile outf = NULL;
AjPStr strand = NULL;
AjPStr substr = NULL;
AjPStr bases = NULL;
ajint begin;
ajint end;
ajint len;
ajint minlen;
float minobsexp;
float minpc;
ajint window;
ajint shift;
ajint plotstart;
ajint plotend;
float *xypc = NULL;
float *obsexp = NULL;
AjBool *thresh = NULL;
float obsexpmax;
ajint i;
ajint maxarr;
embInit("newcpgreport",argc,argv);
seqall = ajAcdGetSeqall("sequence");
window = ajAcdGetInt("window");
shift = ajAcdGetInt("shift");
outf = ajAcdGetOutfile("outfile");
minobsexp = ajAcdGetFloat("minoe");
minlen = ajAcdGetInt("minlen");
minpc = ajAcdGetFloat("minpc");
substr = ajStrNew();
bases = ajStrNewC("CG");
maxarr = 0;
while(ajSeqallNext(seqall, &seq))
{
begin = ajSeqallGetseqBegin(seqall);
end = ajSeqallGetseqEnd(seqall);
strand = ajSeqGetSeqCopyS(seq);
ajStrFmtUpper(&strand);
ajStrAssignSubC(&substr,ajStrGetPtr(strand),--begin,--end);
len=ajStrGetLen(substr);
if(len > maxarr)
{
AJCRESIZE(obsexp, len);
AJCRESIZE(thresh, len);
AJCRESIZE(xypc, len);
maxarr = len;
}
for(i=0;i<len;++i)
obsexp[i]=xypc[i]=0.0;
newcpgreport_findbases(substr, len, window, shift, obsexp,
xypc, bases, &obsexpmax, &plotstart, &plotend);
newcpgreport_identify(outf, obsexp, xypc, thresh, 0, len, shift,
ajStrGetPtr(bases), ajSeqGetNameC(seq), minlen,
minobsexp, minpc, ajStrGetPtr(strand));
ajStrDel(&strand);
}
ajStrDel(&bases);
ajSeqDel(&seq);
ajStrDel(&substr);
ajFileClose(&outf);
AJFREE(obsexp);
AJFREE(thresh);
AJFREE(xypc);
ajSeqallDel(&seqall);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPAlign align;
AjPSeq a;
AjPSeq b;
AjPSeqout seqout;
AjPStr m;
AjPStr n;
AjPStr merged = NULL;
ajuint lena;
ajuint lenb;
const char *p;
const char *q;
ajint start1 = 0;
ajint start2 = 0;
float *path;
ajint *compass;
AjPMatrixf matrix;
AjPSeqCvt cvt = 0;
float **sub;
float gapopen;
float gapextend;
ajulong maxarr = 1000;
ajulong len; /* arbitrary. realloc'd if needed */
size_t stlen;
float score;
ajint begina;
ajint beginb;
embInit("merger", argc, argv);
a = ajAcdGetSeq("asequence");
b = ajAcdGetSeq("bsequence");
seqout = ajAcdGetSeqout("outseq");
matrix = ajAcdGetMatrixf("datafile");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
align = ajAcdGetAlign("outfile");
gapopen = ajRoundFloat(gapopen, 8);
gapextend = ajRoundFloat(gapextend, 8);
AJCNEW(path, maxarr);
AJCNEW(compass, maxarr);
/*
** make the two sequences lowercase so we can show which one we are
** using in the merge by uppercasing it
*/
ajSeqFmtLower(a);
ajSeqFmtLower(b);
m = ajStrNew();
n = ajStrNew();
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
begina = ajSeqGetBegin(a);
beginb = ajSeqGetBegin(b);
lena = ajSeqGetLen(a);
lenb = ajSeqGetLen(b);
if(lenb > (ULONG_MAX/(ajulong)(lena+1)))
ajFatal("Sequences too big. Try 'supermatcher'");
len = lena*lenb;
if(len>maxarr)
{
ajDebug("merger: resize path, len to %d (%d * $d)\n",
len, lena, lenb);
stlen = (size_t) len;
AJCRESIZE(path,stlen);
AJCRESIZE(compass,stlen);
maxarr=len;
}
p = ajSeqGetSeqC(a);
q = ajSeqGetSeqC(b);
ajStrAssignC(&m,"");
ajStrAssignC(&n,"");
score = embAlignPathCalc(p,q,lena,lenb,gapopen,gapextend,path,sub,cvt,
compass, ajFalse);
/*score = embAlignScoreNWMatrix(path,compass,gapopen,gapextend,
a,b,lena,lenb,sub,cvt,
&start1,&start2);*/
embAlignWalkNWMatrix(path,a,b,&m,&n,lena,lenb, &start1,&start2,gapopen,
gapextend,compass);
/*
** now construct the merged sequence, uppercase the bits of the two
** input sequences which are used in the merger
*/
merger_Merge(align, &merged,p,q,m,n,start1,start2,
ajSeqGetNameC(a),ajSeqGetNameC(b));
embAlignReportGlobal(align, a, b, m, n,
start1, start2, gapopen, gapextend,
score, matrix, begina, beginb);
ajAlignWrite(align);
ajAlignReset(align);
/* write the merged sequence */
ajSeqAssignSeqS(a, merged);
ajSeqoutWriteSeq(seqout, a);
ajSeqoutClose(seqout);
ajSeqoutDel(&seqout);
ajSeqDel(&a);
ajSeqDel(&b);
ajAlignClose(align);
ajAlignDel(&align);
ajStrDel(&merged);
AJFREE(compass);
AJFREE(path);
ajStrDel(&n);
ajStrDel(&m);
embExit();
return 0;
}
static AjBool ssematch_NWScore(AjPScop temp_scop,
AjPSeq pseq,
ajint mode,
AjPMatrixf matrix,
float gapopen,
float gapextend)
{
ajint start1 =0; /* Start of seq 1, passed as arg but not used.*/
ajint start2 =0; /* Start of seq 2, passed as arg but not used.*/
ajint maxarr =300; /* Initial size for matrix. */
ajint len;
ajint *compass;
const char *p; /* Query sequence. */
const char *q; /* Subject sequence from scop object. */
float **sub;
float id =0.; /* Passed as arg but not used here. */
float sim =0.;
float idx =0.; /* Passed as arg but not used here. */
float simx =0.; /* Passed as arg but not used here. */
float *path;
AjPStr pstr = NULL; /* m walk alignment for first sequence
Passed as arg but not used here. */
AjPStr qstr = NULL; /* n walk alignment for second sequence
Passed as arg but not used here. */
AjPSeq qseq = NULL; /* Subject sequence. */
ajint lenp; /* Length of query sequence. */
ajint lenq; /* Length of subject sequence. */
AjPSeqCvt cvt = 0;
AjBool show = ajFalse; /*Passed as arg but not used here. */
AJCNEW(path, maxarr);
AJCNEW(compass, maxarr);
pstr = ajStrNew();
qstr = ajStrNew();
gapopen = ajRoundFloat(gapopen,8);
gapextend = ajRoundFloat(gapextend,8);
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
/* Extract subject sequence from scop object, convert to 3 letter code. */
if (mode == 0)
qseq = ssematch_convertbases(temp_scop->Sse);
else if (mode == 1)
qseq = ssematch_convertbases(temp_scop->Sss);
lenp = ajSeqGetLen(pseq); /* Length of query sequence. */
lenq = ajSeqGetLen(qseq); /* Length of subject sequence. */
/* Start of main application loop */
/* Intitialise variables for use by alignment functions*/
len = (lenp * lenq);
if(len>maxarr)
{
AJCRESIZE(path,len);
AJCRESIZE(compass,len);
maxarr=len;
}
p = ajSeqGetSeqC(pseq);
q = ajSeqGetSeqC(qseq);
ajStrAssignC(&pstr,"");
ajStrAssignC(&qstr,"");
/* Check that no sequence length is 0. */
if((lenp == 0)||(lenq == 0))
{
AJFREE(compass);
AJFREE(path);
ajStrDel(&pstr);
ajStrDel(&qstr);
}
/* Call alignment functions. */
embAlignPathCalc(p,q,lenp,lenq, gapopen,
gapextend,path,sub,cvt,compass,show);
/*embAlignScoreNWMatrix(path,compass,gapopen,gapextend,
pseq, qseq,
lenp,lenq,sub,cvt,
&start1,&start2);*/
embAlignWalkNWMatrix(path,pseq,qseq,&pstr,&qstr,
lenp,lenq,&start1,&start2,
gapopen,gapextend,compass);
embAlignCalcSimilarity(pstr,qstr,sub,cvt,lenp,
lenq,&id,&sim,&idx, &simx);
/* Assign score. */
temp_scop->Score = sim;
/* Tidy up */
AJFREE(compass);
AJFREE(path);
ajStrDel(&pstr);
ajStrDel(&qstr);
ajSeqDel(&qseq);
/* Bye Bye */
return ajTrue;
}
/* @funcstatic newcoils_read_matrix *******************************************
**
** Reads the matrix and stores in a hept_pref structure
**
** @param [u] inf [AjPFile] matrix input file
** @return [struct hept_pref*] Matrix data for heptad preference
******************************************************************************/
static struct hept_pref* newcoils_read_matrix(AjPFile inf)
{
ajint i;
ajint j;
ajint pt;
ajint aa_len;
ajint win;
float m_g;
float sd_g;
float m_cc;
float sd_cc;
float sc;
float hept[NCHEPTAD];
AjPStr buff;
const char *pbuff;
struct hept_pref *h;
buff = ajStrNew();
aa_len = strlen(NCAAs);
AJNEW(h);
AJCNEW(h->m,aa_len);
for(i=0; i<aa_len; ++i)
{
AJCNEW(h->m[i],NCHEPTAD);
for(j=0; j<NCHEPTAD; ++j)
h->m[i][j] = -1;
}
AJNEW(h->f);
h->n = 0;
h->smallest = 1.0;
while(ajReadlineTrim(inf,&buff))
{
pbuff = ajStrGetPtr(buff);
if(*pbuff != '%')
{
if((strncmp(pbuff,"uw ",3)==0) || (strncmp(pbuff,"w ",2)==0))
{
i = h->n;
if(strncmp(pbuff,"uw ",3)==0)
h->f[i].w = 0;
else
h->f[i].w = 1;
ajFmtScanS(buff,"%*s %d %f %f %f %f %f",&win,&m_cc,
&sd_cc,&m_g,&sd_g,&sc);
h->f[i].win = win;
h->f[i].m_cc = (float)m_cc;
h->f[i].sd_cc = (float)sd_cc;
h->f[i].m_g = (float)m_g;
h->f[i].sd_g = (float)sd_g;
h->f[i].sc = (float)sc;
h->n++;
AJCRESIZE(h->f,(h->n)+1);
if((h->n)>=9)
ajFatal("Too many window parms in matrix file\n");
}
else if(*pbuff>='A' && *pbuff<='Z')
{
/* AA data */
pt = (int)(pbuff[0]-'A');
if(h->m[pt][0]==-1)
{
ajFmtScanS(buff,"%*s%f%f%f%f%f%f%f",&hept[0],
&hept[1],&hept[2],&hept[3],&hept[4],
&hept[5],&hept[6]);
for(i=0; i<NCHEPTAD; ++i)
{
h->m[pt][i] = (float)hept[i];
if(h->m[pt][i]>0)
{
if(h->m[pt][i]<h->smallest)
h->smallest = h->m[pt][i];
}
else
h->m[pt][i]=-1; /* Don't permit zero values */
}
}
else
ajWarn("multiple entries for AA %c in matrix file\n",
*pbuff);
}
else
{
ajWarn("strange characters in matrix file\n");
ajWarn("Ignoring line: %S\n",buff);
}
}
}
ajStrDel(&buff);
return h;
}
int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq = NULL;
AjPFile inf = NULL;
AjPStr strand = NULL;
AjPStr substr = NULL;
AjPStr name = NULL;
AjPStr mname = NULL;
AjPStr tname = NULL;
AjPStr pname = NULL;
AjPStr line = NULL;
AjPStr cons = NULL;
AjPStr m = NULL;
AjPStr n = NULL;
AjPAlign align= NULL; /* JISON, replaces AjPOutfile outf */
ajint type;
ajint begin;
ajint end;
ajulong len;
ajint i;
ajint j;
float **fmatrix=NULL;
ajint mlen;
float maxfs;
ajint thresh;
float gapopen;
float gapextend;
float opencoeff;
float extendcoeff;
const char *p;
ajulong maxarr = 1000;
ajulong alen;
float *path;
ajint *compass;
size_t stlen;
embInit("prophet", argc, argv);
seqall = ajAcdGetSeqall("sequence");
inf = ajAcdGetInfile("infile");
opencoeff = ajAcdGetFloat("gapopen");
extendcoeff = ajAcdGetFloat("gapextend");
align = ajAcdGetAlign("outfile"); /*JISON replacing outfile */
opencoeff = ajRoundFloat(opencoeff, 8);
extendcoeff = ajRoundFloat(extendcoeff, 8);
substr = ajStrNew();
name = ajStrNew();
mname = ajStrNew();
tname = ajStrNew();
line = ajStrNew();
m = ajStrNewC("");
n = ajStrNewC("");
type = prophet_getType(inf,&tname);
if(!type)
ajFatal("Unrecognised profile/matrix file format");
prophet_read_profile(inf,&pname,&mname,&mlen,&gapopen,&gapextend,&thresh,
&maxfs, &cons);
ajAlignSetMatrixName(align, mname);
AJCNEW(fmatrix, mlen);
for(i=0;i<mlen;++i)
{
AJCNEW(fmatrix[i], AZ);
if(!ajReadlineTrim(inf,&line))
ajFatal("Missing matrix line");
p = ajStrGetPtr(line);
p = ajSysFuncStrtok(p," \t");
for(j=0;j<AZ;++j)
{
sscanf(p,"%f",&fmatrix[i][j]);
p = ajSysFuncStrtok(NULL," \t");
}
}
AJCNEW(path, maxarr);
AJCNEW(compass, maxarr);
while(ajSeqallNext(seqall, &seq))
{
begin = ajSeqallGetseqBegin(seqall);
end = ajSeqallGetseqEnd(seqall);
ajStrAssignC(&name,ajSeqGetNameC(seq));
strand = ajSeqGetSeqCopyS(seq);
ajStrAssignSubC(&substr,ajStrGetPtr(strand),begin-1,end-1);
len = ajStrGetLen(substr);
if(len > (ULONG_MAX/(ajulong)(mlen+1)))
ajFatal("Sequences too big. Try 'supermatcher'");
alen = len*mlen;
if(alen>maxarr)
{
stlen = (size_t) alen;
AJCRESIZE(path,stlen);
AJCRESIZE(compass,stlen);
maxarr=alen;
}
ajStrAssignC(&m,"");
ajStrAssignC(&n,"");
/* JISON used to be
prophet_scan_profile(substr,pname,name,mlen,fmatrix,
outf,cons,opencoeff,
extendcoeff,path,compass,&m,&n,len); */
/* JISON new call and reset align */
prophet_scan_profile(substr,name,pname,mlen,fmatrix,
align,cons,opencoeff,
extendcoeff,path,compass,&m,&n,(ajint)len);
ajAlignReset(align);
ajStrDel(&strand);
}
for(i=0;i<mlen;++i)
AJFREE (fmatrix[i]);
AJFREE (fmatrix);
AJFREE(path);
AJFREE(compass);
ajStrDel(&line);
ajStrDel(&cons);
ajStrDel(&name);
ajStrDel(&pname);
ajStrDel(&mname);
ajStrDel(&tname);
ajStrDel(&substr);
ajStrDel(&m);
ajStrDel(&n);
ajSeqDel(&seq);
ajFileClose(&inf);
ajAlignClose(align);
ajAlignDel(&align);
ajSeqallDel(&seqall);
embExit();
return 0;
}
