int main(int argc, char **argv)
{
/* ACD data item variables */
AjPSeqset dataset = NULL;
AjPFile bfile = NULL;
AjPFile plib = NULL;
AjPStr mod = NULL;
ajint nmotifs = 0;
AjBool text = ajFalse;
AjPStr prior = NULL;
float evt = 0.0;
ajint nsites = 0;
ajint minsites = 0;
ajint maxsites = 0;
float wnsites = 0.0;
ajint w = 0;
ajint minw = 0;
ajint maxw = 0;
AjBool nomatrim = ajFalse;
ajint wg = 0;
ajint ws = 0;
AjBool noendgaps = ajFalse;
AjBool revcomp = ajFalse;
AjBool pal = ajFalse;
AjBool nostatus = ajFalse;
ajint maxiter = 0;
float distance = 0.0;
float b = 0.0;
float spfuzz = 0.0;
AjPStr spmap = NULL;
AjPStr cons = NULL;
ajint maxsize = 0;
ajint p = 0;
ajint time = 0;
AjPStr sf = NULL;
ajint heapsize = 64;
AjBool xbranch = ajFalse;
AjBool wbranch = ajFalse;
ajint bfactor = 0;
AjPFile outtext = NULL;
/* Housekeeping variables */
AjPStr cmd = NULL;
AjPStr ssname = NULL;
AjPSeqout outseq = NULL;
AjPStr tmp = NULL;
char option;
/* ACD file processing */
embInitPV("ememetext",argc,argv,"MEME",VERSION);
dataset = ajAcdGetSeqset("dataset");
bfile = ajAcdGetInfile("bfile");
plib = ajAcdGetInfile("plibfile");
mod = ajAcdGetSelectSingle("mod");
nmotifs = ajAcdGetInt("nmotifs");
text = ajAcdGetBoolean("text");
prior = ajAcdGetSelectSingle("prior");
evt = ajAcdGetFloat("evt");
nsites = ajAcdGetInt("nsites");
minsites = ajAcdGetInt("minsites");
maxsites = ajAcdGetInt("maxsites");
wnsites = ajAcdGetFloat("wnsites");
w = ajAcdGetInt("w");
minw = ajAcdGetInt("minw");
maxw = ajAcdGetInt("maxw");
nomatrim = ajAcdGetBoolean("nomatrim");
wg = ajAcdGetInt("wg");
ws = ajAcdGetInt("ws");
noendgaps = ajAcdGetBoolean("noendgaps");
revcomp = ajAcdGetBoolean("revcomp");
pal = ajAcdGetBoolean("pal");
nostatus = ajAcdGetBoolean("nostatus");
maxiter = ajAcdGetInt("maxiter");
distance = ajAcdGetFloat("distance");
b = ajAcdGetFloat("b");
spfuzz = ajAcdGetFloat("spfuzz");
spmap = ajAcdGetSelectSingle("spmap");
cons = ajAcdGetString("cons");
maxsize = ajAcdGetInt("maxsize");
p = ajAcdGetInt("p");
time = ajAcdGetInt("time");
sf = ajAcdGetString("sf");
heapsize = ajAcdGetInt("heapsize");
xbranch = ajAcdGetBoolean("xbranch");
wbranch = ajAcdGetBoolean("wbranch");
bfactor = ajAcdGetInt("bfactor");
outtext = ajAcdGetOutfile("outtext");
outseq = ajAcdGetSeqoutset("outseq");
/* MAIN APPLICATION CODE */
/* 1. Housekeeping */
cmd = ajStrNew();
tmp = ajStrNew();
/* 2. Re-write dataset to a temporary file in a format (fasta) MEME
** can understand.
** Can't just pass the name of dataset to MEME as the name provided
** might be a USA which MEME would not understand.
*/
ssname = ajStrNewS(ajFileGetNameS(outseq->File));
ajSeqoutSetFormatC(outseq, "fasta");
ajSeqoutWriteSet(outseq, dataset);
ajSeqoutClose(outseq);
ajSeqoutDel(&outseq);
/* 3. Build ememe command line */
/* Command line is built in this order:
i. Application name.
ii. Original MEME options (in order they appear in ACD file)
iii.Original MEME options (that don't appear in ACD file)
iv. EMBASSY MEME new qualifiers and parameters.
*/
ajStrAssignS(&cmd, ajAcdGetpathC("meme"));
ajFmtPrintAppS(&cmd, " %S", ssname);
if(bfile)
ajFmtPrintAppS(&cmd, " -bfile %s ", ajFileGetNameC(bfile));
if(plib)
ajFmtPrintAppS(&cmd, " -plib %s ", ajFileGetNameC(plib));
option = ajStrGetCharFirst(mod);
if(option == 'o')
ajStrAppendC(&cmd, " -mod oops ");
else if(option == 'z')
ajStrAppendC(&cmd, " -mod zoops ");
else if(option == 'a')
ajStrAppendC(&cmd, " -mod anr ");
if(nmotifs != 1)
ajFmtPrintAppS(&cmd, " -nmotifs %d ", nmotifs);
if(text)
ajFmtPrintAppS(&cmd, " -text ");
ajFmtPrintAppS(&cmd, " -prior %S ", prior);
if(evt != -1)
ajFmtPrintAppS(&cmd, " -evt %f ", evt);
if(nsites != -1)
ajFmtPrintAppS(&cmd, " -nsites %d ", nsites);
else
{
if(minsites != -1)
ajFmtPrintAppS(&cmd, " -minsites %d ", minsites);
if(maxsites != -1)
ajFmtPrintAppS(&cmd, " -maxsites %d ", maxsites);
}
if(wnsites < 0.7999 || wnsites > .8001)
ajFmtPrintAppS(&cmd, " -wnsites %f ", wnsites);
if(w != -1)
ajFmtPrintAppS(&cmd, " -w %d ", w);
if(minw != 8)
ajFmtPrintAppS(&cmd, " -minw %d ", minw);
if(maxw != 50)
ajFmtPrintAppS(&cmd, " -maxw %d ", maxw);
if(nomatrim)
ajFmtPrintAppS(&cmd, " -nomatrim ");
if(wg != 11)
ajFmtPrintAppS(&cmd, " -wg %d ", wg);
if(ws != 1)
ajFmtPrintAppS(&cmd, " -ws %d ", ws);
if(noendgaps)
ajFmtPrintAppS(&cmd, " -noendgaps ");
if(revcomp)
ajFmtPrintAppS(&cmd, " -revcomp ");
if(pal && ajSeqsetIsNuc(dataset))
ajFmtPrintAppS(&cmd, " -pal ");
if(nostatus)
ajFmtPrintAppS(&cmd, " -nostatus ");
if(maxiter != 50)
ajFmtPrintAppS(&cmd, " -maxiter %d ", maxiter);
if(distance < 0.00099 || distance > 0.00101)
ajFmtPrintAppS(&cmd, " -distance %f ", distance);
if(b != -1)
ajFmtPrintAppS(&cmd, " -b %f ", b);
if(spfuzz != -1)
ajFmtPrintAppS(&cmd, " -spfuzz %f ", spfuzz);
if(!ajStrMatchC(spmap,"default"))
ajFmtPrintAppS(&cmd, " -spmap %S ", spmap);
if(MAJSTRGETLEN(cons))
ajFmtPrintAppS(&cmd, "-cons %S", cons);
if(maxsize != -1)
ajFmtPrintAppS(&cmd, " -maxsize %d ", maxsize);
if(p > 0)
ajFmtPrintAppS(&cmd, " -p %d ", p);
if(time > 0)
ajFmtPrintAppS(&cmd, " -time %d ", time);
if(MAJSTRGETLEN(sf))
ajFmtPrintAppS(&cmd, " -sf %S", sf);
if(heapsize != 64)
ajFmtPrintAppS(&cmd, " -heapsize %d ", heapsize);
if(xbranch)
ajFmtPrintAppS(&cmd, " -x_branch");
if(wbranch)
ajFmtPrintAppS(&cmd, " -w_branch");
if(bfactor != 3)
ajFmtPrintAppS(&cmd, " -bfactor %d ", bfactor);
if(ajSeqsetIsProt(dataset))
ajFmtPrintAppS(&cmd, "-protein ");
else
ajFmtPrintAppS(&cmd, "-dna ");
ajFmtPrintAppS(&cmd, " -text");
ajFmtPrintAppS(&cmd, " > %S ", ajFileGetNameS(outtext));
/* 4. Close files from ACD before calling meme */
ajFileClose(&bfile);
ajFileClose(&plib);
/* 5. Call meme */
/* ajFmtPrint("\n%S\n", cmd); */
system(ajStrGetPtr(cmd));
/* 6. Exit cleanly */
ajSeqsetDel(&dataset);
ajStrDel(&cons);
ajStrDel(&sf);
ajStrDel(&mod);
ajStrDel(&prior);
ajStrDel(&spmap);
ajStrDel(&cmd);
ajStrDel(&ssname);
ajStrDel(&tmp);
ajFileClose(&bfile);
ajFileClose(&plib);
ajFileClose(&outtext);
ajSeqoutDel(&outseq);
embExit();
return 0;
}
void embConsCalc(const AjPSeqset seqset,const AjPMatrix cmpmatrix,
ajint nseqs, ajint mlen,float fplural,float setcase,
ajint identity, AjBool gaps, AjPStr *cons)
{
ajint i;
ajint j;
ajint k;
ajint **matrix;
ajint m1 = 0;
ajint m2 = 0;
ajint highindex;
ajint matsize;
ajint matchingmaxindex;
ajint identicalmaxindex;
float max;
float contri = 0;
float contrj = 0;
float *identical;
float *matching;
AjPSeqCvt cvt = 0;
AjPFloat score = NULL;
const char **seqcharptr;
char res;
char nocon = '-';
void *freeptr;
matrix = ajMatrixGetMatrix(cmpmatrix);
cvt = ajMatrixGetCvt(cmpmatrix); /* return conversion table */
matsize = ajMatrixGetSize(cmpmatrix);
AJCNEW(seqcharptr,nseqs);
AJCNEW(identical,matsize);
AJCNEW(matching,matsize);
score = ajFloatNew();
if(ajSeqsetIsNuc(seqset)) /* set non-consensus character */
nocon = 'N';
else if ( ajSeqsetIsProt(seqset))
nocon = 'X';
for(i=0;i<nseqs;i++) /* get sequence as string */
seqcharptr[i] = ajSeqsetGetseqSeqC(seqset, i);
for(k=0; k< mlen; k++)
{
res = nocon;
for(i=0;i<matsize;i++) /* reset id's and +ve matches */
{
identical[i] = 0.0;
matching[i] = 0.0;
}
for(i=0;i<nseqs;i++)
ajFloatPut(&score,i,0.);
for(i=0;i<nseqs;i++) /* generate score for columns */
{
m1 = ajSeqcvtGetCodeK(cvt,seqcharptr[i][k]);
if(m1 || gaps)
identical[m1] += ajSeqsetGetseqWeight(seqset,i);
for(j=i+1;j<nseqs;j++)
{
m2 = ajSeqcvtGetCodeK(cvt,seqcharptr[j][k]);
if(m1 && m2)
{
contri = (float)matrix[m1][m2]*
ajSeqsetGetseqWeight(seqset,j)
+ajFloatGet(score,i);
contrj = (float)matrix[m1][m2]*
ajSeqsetGetseqWeight(seqset,i)
+ajFloatGet(score,j);
ajFloatPut(&score,i,contri);
ajFloatPut(&score,j,contrj);
}
}
}
highindex = -1;
max = -(float)INT_MAX;
for(i=0;i<nseqs;i++)
if( ajFloatGet(score,i) > max ||
(ajFloatGet(score,i) == max &&
seqcharptr[highindex][k] == '-') )
{
highindex = i;
max = ajFloatGet(score,i);
}
for(i=0;i<nseqs;i++) /* find +ve matches in the column */
{
m1 = ajSeqcvtGetCodeK(cvt, seqcharptr[i][k]);
if(!matching[m1])
for(j=0;j<nseqs;j++)
{
/*
// if( i != j)
// {
// m2 = ajSeqcvtGetCodeK(cvt, seqcharptr[j][k]);
// if(m1 && m2 && matrix[m1][m2] > 0)
// matching[m1] += ajSeqsetGetseqWeight(seqset, j);
// }
*/
m2 = ajSeqcvtGetCodeK(cvt, seqcharptr[j][k]);
if(m1 && m2 && matrix[m1][m2] > 0)
matching[m1] += ajSeqsetGetseqWeight(seqset, j);
if(gaps && !m1 && !m2)
matching[m1] += ajSeqsetGetseqWeight(seqset, j);
}
}
matchingmaxindex = 0; /* get max matching and identical */
identicalmaxindex = 0;
for(i=0;i<nseqs;i++)
{
m1 = ajSeqcvtGetCodeK(cvt,seqcharptr[i][k]);
if(identical[m1] > identical[identicalmaxindex])
identicalmaxindex = m1;
}
for(i=0;i<nseqs;i++)
{
m1 = ajSeqcvtGetCodeK(cvt,seqcharptr[i][k]);
if(matching[m1] > matching[matchingmaxindex])
matchingmaxindex = m1;
else if(matching[m1] == matching[matchingmaxindex])
if(identical[m1] > identical[matchingmaxindex])
matchingmaxindex = m1;
}
/* plurality check */
m1 = ajSeqcvtGetCodeK(cvt,seqcharptr[highindex][k]);
/* if(matching[m1] >= fplural
&& seqcharptr[highindex][k] != '-')
res = seqcharptr[highindex][k];*/
if(matching[m1] >= fplural)
res = seqcharptr[highindex][k];
if(matching[m1]<= setcase)
res = tolower((int)res);
if(identity) /* if just looking for id's */
{
j = 0;
for(i=0;i<nseqs;i++)
if(matchingmaxindex == ajSeqcvtGetCodeK(cvt,seqcharptr[i][k]))
j++;
if(j<identity)
res = nocon;
}
ajStrAppendK(cons,res);
}
freeptr = (void *) seqcharptr;
AJFREE(freeptr);
AJFREE(matching);
AJFREE(identical);
ajFloatDel(&score);
return;
}