int main(int argc, char **argv)
{
AjPSeqall seqall;
AjPSeq seq;
AjPTable table = 0;
AjPFile outf;
ajint wordsize;
ajint mincount;
embInit("wordcount", argc, argv);
seqall = ajAcdGetSeqall("sequence1");
wordsize = ajAcdGetInt("wordsize");
outf = ajAcdGetOutfile("outfile");
mincount = ajAcdGetInt("mincount");
embWordLength(wordsize);
while (ajSeqallNext(seqall, &seq))
{
embWordGetTable(&table, seq); /* get table of words */
}
embWordPrintTableFI(table, mincount, outf); /* print table of words */
/*
** test if table can be added to
** if(getWordTable(&table, seq, wordcount)) ?? get table of words ??
** {
** printWordTable(table); ?? print table of words ??
** }
*/
embWordFreeTable(&table); /* free table of words */
ajSeqallDel(&seqall);
ajSeqDel(&seq);
ajFileClose(&outf);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall seq1;
AjPSeqset seq2;
AjPSeq a;
const AjPSeq b;
AjPStr m = 0;
AjPStr n = 0;
AjPFile errorf;
AjBool show = ajFalse;
ajint lena = 0;
ajint lenb = 0;
const char *p;
const char *q;
AjPMatrixf matrix;
AjPSeqCvt cvt = 0;
float **sub;
ajint *compass = NULL;
float *path = NULL;
float gapopen;
float gapextend;
float score;
ajint begina;
ajint i;
ajuint k;
ajint beginb;
ajint start1 = 0;
ajint start2 = 0;
ajint end1 = 0;
ajint end2 = 0;
ajint width = 0;
AjPTable seq1MatchTable = 0;
ajint wordlen = 6;
ajint oldmax = 0;
AjPAlign align = NULL;
embInit("supermatcher", argc, argv);
matrix = ajAcdGetMatrixf("datafile");
seq1 = ajAcdGetSeqall("asequence");
seq2 = ajAcdGetSeqset("bsequence");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
wordlen = ajAcdGetInt("wordlen");
align = ajAcdGetAlign("outfile");
errorf = ajAcdGetOutfile("errorfile");
width = ajAcdGetInt("width"); /* not the same as awidth */
gapopen = ajRoundFloat(gapopen, 8);
gapextend = ajRoundFloat(gapextend, 8);
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
embWordLength(wordlen);
ajSeqsetTrim(seq2);
while(ajSeqallNext(seq1,&a))
{
ajSeqTrim(a);
begina = 1 + ajSeqGetOffset(a);
m = ajStrNewRes(1+ajSeqGetLen(a));
lena = ajSeqGetLen(a);
ajDebug("Read '%S'\n", ajSeqGetNameS(a));
if(!embWordGetTable(&seq1MatchTable, a)) /* get table of words */
ajErr("Could not generate table for %s\n",
ajSeqGetNameC(a));
for(k=0;k<ajSeqsetGetSize(seq2);k++)
{
b = ajSeqsetGetseqSeq(seq2, k);
lenb = ajSeqGetLen(b);
beginb = 1 + ajSeqGetOffset(b);
ajDebug("Processing '%S'\n", ajSeqGetNameS(b));
p = ajSeqGetSeqC(a);
q = ajSeqGetSeqC(b);
if(!supermatcher_findstartpoints(seq1MatchTable,b,a,
&start1, &start2,
&end1, &end2))
{
ajFmtPrintF(errorf,
"No wordmatch start points for "
"%s vs %s. No alignment\n",
ajSeqGetNameC(a),ajSeqGetNameC(b));
continue;
}
n=ajStrNewRes(1+ajSeqGetLen(b));
ajStrAssignC(&m,"");
ajStrAssignC(&n,"");
ajDebug("++ %S v %S start:%d %d end:%d %d\n",
ajSeqGetNameS(a), ajSeqGetNameS(b),
start1, start2, end1, end2);
if(end1-start1+1 > oldmax)
{
oldmax = ((end1-start1)+1);
AJRESIZE(path,oldmax*width*sizeof(float));
AJRESIZE(compass,oldmax*width*sizeof(ajint));
ajDebug("++ resize to oldmax: %d\n", oldmax);
}
for(i=0;i<((end1-start1)+1)*width;i++)
path[i] = 0.0;
ajDebug("Calling embAlignPathCalcFast "
"%d..%d [%d/%d] %d..%d [%d/%d]\n",
start1, end1, (end1 - start1 + 1), lena,
start2, end2, (end2 - start2 + 1), lenb);
score = embAlignPathCalcSWFast(&p[start1],&q[start2],
end1-start1+1,end2-start2+1,
0,width,
gapopen,gapextend,
path,sub,cvt,
compass,show);
embAlignWalkSWMatrixFast(path,compass,gapopen,gapextend,a,b,
&m,&n,end1-start1+1,end2-start2+1,
0,width,
&start1,&start2);
if(!ajAlignFormatShowsSequences(align))
{
ajAlignDefineCC(align, ajStrGetPtr(m),
ajStrGetPtr(n), ajSeqGetNameC(a),
ajSeqGetNameC(b));
ajAlignSetScoreR(align, score);
}
else
{
embAlignReportLocal(align, a, b,
m,n,start1,start2,
gapopen, gapextend,
score,matrix, begina, beginb);
}
ajAlignWrite(align);
ajAlignReset(align);
ajStrDel(&n);
}
embWordFreeTable(&seq1MatchTable); /* free table of words */
seq1MatchTable=0;
ajStrDel(&m);
}
if(!ajAlignFormatShowsSequences(align))
{
ajMatrixfDel(&matrix);
}
AJFREE(path);
AJFREE(compass);
ajAlignClose(align);
ajAlignDel(&align);
ajSeqallDel(&seq1);
ajSeqDel(&a);
ajSeqsetDel(&seq2);
ajFileClose(&errorf);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeq seq1;
AjPSeq seq2;
ajint wordlen;
AjPTable seq1MatchTable = 0;
AjPList matchlist = NULL;
AjPGraph graph = NULL;
AjPGraph xygraph = NULL;
AjBool boxit;
/*
** Different ticks as they need to be different for x and y due to
** length of string being important on x
*/
ajuint acceptableticksx[]=
{
1,10,50,100,500,1000,1500,10000,
500000,1000000,5000000
};
ajuint acceptableticks[]=
{
1,10,50,100,200,500,1000,2000,5000,10000,15000,
500000,1000000,5000000
};
ajint numbofticks = 10;
float xmargin;
float ymargin;
float ticklen;
float tickgap;
float onefifth = 0.0;
ajint i;
float k;
float max;
char ptr[10];
ajint begin1;
ajint begin2;
ajint end1;
ajint end2;
ajuint len1;
ajuint len2;
float fbegin1;
float fbegin2;
float fend1;
float fend2;
float flen1;
float flen2;
AjBool stretch;
embInit("dottup", argc, argv);
wordlen = ajAcdGetInt("wordsize");
seq1 = ajAcdGetSeq("asequence");
seq2 = ajAcdGetSeq("bsequence");
graph = ajAcdGetGraph("graph");
boxit = ajAcdGetBoolean("boxit");
stretch = ajAcdGetToggle("stretch");
xygraph = ajAcdGetGraphxy("xygraph");
begin1 = ajSeqGetBegin(seq1);
begin2 = ajSeqGetBegin(seq2);
end1 = ajSeqGetEnd(seq1);
end2 = ajSeqGetEnd(seq2);
len1 = end1 - begin1 + 1;
len2 = end2 - begin2 + 1;
flen1 = (float) len1;
flen2 = (float) len2;
fbegin1 = (float) begin1;
fbegin2 = (float) begin2;
fend1 = (float) end1;
fend2 = (float) end2;
offset1 = fbegin1;
offset2 = fbegin2;
ajSeqTrim(seq1);
ajSeqTrim(seq2);
embWordLength(wordlen);
if(embWordGetTable(&seq1MatchTable, seq1))
matchlist = embWordBuildMatchTable(seq1MatchTable, seq2, ajTrue);
if(stretch)
{
dottup_stretchplot(xygraph,matchlist,seq1,seq2,begin1,begin2,end1,
end2);
if(matchlist)
embWordMatchListDelete(&matchlist); /* free the match structures */
}
else
{
/* only here if stretch is false */
max= flen1;
if(flen2 > max)
max = flen2;
xmargin = ymargin = max * (float)0.15;
ajGraphOpenWin(graph, fbegin1-ymargin,fend1+ymargin,
fbegin2-xmargin,(float)fend2+xmargin);
ajGraphicsSetCharscale(0.5);
if(matchlist)
dottup_plotMatches(matchlist);
if(boxit)
{
ajGraphicsDrawposRect(fbegin1, fbegin2, fend1, fend2);
i = 0;
while(acceptableticksx[i]*numbofticks < len1)
i++;
if(i<=13)
tickgap = (float) acceptableticksx[i];
else
tickgap = (float) acceptableticksx[10];
ticklen = xmargin * (float) 0.1;
onefifth = xmargin * (float)0.2;
ajGraphicsDrawposTextAtmid(fbegin1+flen1*(float)0.5,
fbegin1-(onefifth*(float)3.0),
ajGraphGetYlabelC(graph));
if(len2/len1 > 10 )
{
/* a lot smaller then just label start and end */
ajGraphicsDrawposLine(fbegin1,fbegin2,fbegin1,
fbegin2-ticklen);
sprintf(ptr,"%u",ajSeqGetOffset(seq1));
ajGraphicsDrawposTextAtmid(fbegin1,fbegin2-(onefifth),ptr);
ajGraphicsDrawposLine(fend1,fbegin2,
fend1,fbegin2-ticklen);
sprintf(ptr,"%d",end1);
ajGraphicsDrawposTextAtmid(fend1,fbegin2-(onefifth),ptr);
}
else
for(k=fbegin1;k<fend1;k+=tickgap)
{
ajGraphicsDrawposLine(k,fbegin2,k,fbegin2-ticklen);
sprintf(ptr,"%d",(ajint)k);
ajGraphicsDrawposTextAtmid( k,fbegin2-(onefifth),ptr);
}
i = 0;
while(acceptableticks[i]*numbofticks < len2)
i++;
tickgap = (float) acceptableticks[i];
ticklen = ymargin*(float)0.1;
onefifth = ymargin*(float)0.2;
ajGraphicsDrawposTextAtlineJustify(fbegin1-(onefifth*(float)4.),
fbegin2+flen2*(float)0.5,
fbegin2-(onefifth*(float)4.),
fbegin2+flen2,
ajGraphGetXlabelC(graph),
0.5);
if(len1/len2 > 10 )
{
/* a lot smaller then just label start and end */
ajGraphicsDrawposLine(fbegin1,fbegin2,fbegin1-ticklen,
fbegin2);
sprintf(ptr,"%u",ajSeqGetOffset(seq2));
ajGraphicsDrawposTextAtend(fbegin1-(onefifth),fbegin2,ptr);
ajGraphicsDrawposLine(fbegin1,fend2,fbegin1-ticklen,
fend2);
sprintf(ptr,"%d",end2);
ajGraphicsDrawposTextAtend(fbegin2-(onefifth),fend2,ptr);
}
else
for(k=fbegin2;k<fend2;k+=tickgap)
{
ajGraphicsDrawposLine(fbegin1,k,fbegin1-ticklen,k);
sprintf(ptr,"%d",(ajint)k);
ajGraphicsDrawposTextAtend(fbegin1-(onefifth),k,ptr);
}
}
}
ajGraphicsClose();
ajSeqDel(&seq1);
ajSeqDel(&seq2);
ajGraphxyDel(&graph);
ajGraphxyDel(&xygraph);
embWordFreeTable(&seq1MatchTable);
if(matchlist)
embWordMatchListDelete(&matchlist); /* free the match structures */
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset;
const AjPSeq seq1;
const AjPSeq seq2;
ajint wordlen;
AjPTable seq1MatchTable = NULL;
AjPList matchlist ;
AjPGraph graph = 0;
ajuint i;
ajuint j;
float total=0;
ajuint acceptableticks[]=
{
1,10,50,100,200,500,1000,1500,10000,50000,
100000,500000,1000000,5000000
};
ajint numbofticks = 10;
ajint gap,tickgap;
AjBool boxit = AJTRUE;
AjBool dumpfeat = AJFALSE;
float xmargin;
float ymargin;
float k;
char ptr[10];
float ticklen;
float onefifth;
AjPFeattable *tabptr = NULL;
AjPFeattabOut seq1out = NULL;
AjPStr sajb = NULL;
float flen1;
float flen2;
ajuint tui;
embInit("polydot", argc, argv);
wordlen = ajAcdGetInt("wordsize");
seqset = ajAcdGetSeqset("sequences");
graph = ajAcdGetGraph("graph");
gap = ajAcdGetInt("gap");
boxit = ajAcdGetBoolean("boxit");
seq1out = ajAcdGetFeatout("outfeat");
dumpfeat = ajAcdGetToggle("dumpfeat");
sajb = ajStrNew();
embWordLength(wordlen);
AJCNEW(lines,ajSeqsetGetSize(seqset));
AJCNEW(pts,ajSeqsetGetSize(seqset));
AJCNEW(tabptr,ajSeqsetGetSize(seqset));
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
seq1 = ajSeqsetGetseqSeq(seqset, i);
total += ajSeqGetLen(seq1);
}
total +=(float)(gap*(ajSeqsetGetSize(seqset)-1));
xmargin = total*(float)0.15;
ymargin = total*(float)0.15;
ticklen = xmargin*(float)0.1;
onefifth = xmargin*(float)0.2;
i = 0;
while(acceptableticks[i]*numbofticks < ajSeqsetGetLen(seqset))
i++;
if(i<=13)
tickgap = acceptableticks[i];
else
tickgap = acceptableticks[13];
ajGraphAppendTitleS(graph, ajSeqsetGetUsa(seqset));
ajGraphOpenWin(graph, (float)0.0-xmargin,(total+xmargin)*(float)1.35,
(float)0.0-ymargin,
total+ymargin);
ajGraphicsSetCharscale((float)0.3);
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
which = i;
seq1 = ajSeqsetGetseqSeq(seqset, i);
tui = ajSeqGetLen(seq1);
flen1 = (float) tui;
if(embWordGetTable(&seq1MatchTable, seq1)){ /* get table of words */
for(j=0;j<ajSeqsetGetSize(seqset);j++)
{
seq2 = ajSeqsetGetseqSeq(seqset, j);
tui = ajSeqGetLen(seq2);
flen2 = (float) tui;
if(boxit)
ajGraphicsDrawposRect(xstart,ystart,
xstart+flen1,
ystart+flen2);
matchlist = embWordBuildMatchTable(seq1MatchTable, seq2,
ajTrue);
if(matchlist)
polydot_plotMatches(matchlist);
if(i<j && dumpfeat)
embWordMatchListConvToFeat(matchlist,&tabptr[i],
&tabptr[j],seq1,
seq2);
if(matchlist) /* free the match structures */
embWordMatchListDelete(&matchlist);
if(j==0)
{
for(k=0.0;k<ajSeqGetLen(seq1);k+=tickgap)
{
ajGraphicsDrawposLine(xstart+k,ystart,xstart+k,
ystart-ticklen);
sprintf(ptr,"%d",(ajint)k);
ajGraphicsDrawposTextAtmid(xstart+k,
ystart-(onefifth),
ptr);
}
ajGraphicsDrawposTextAtmid(
xstart+(flen1/(float)2.0),
ystart-(3*onefifth),
ajStrGetPtr(ajSeqsetGetseqNameS(seqset, i)));
}
if(i==0)
{
for(k=0.0;k<ajSeqGetLen(seq2);k+=tickgap)
{
ajGraphicsDrawposLine(xstart,ystart+k,xstart-ticklen,
ystart+k);
sprintf(ptr,"%d",(ajint)k);
ajGraphicsDrawposTextAtend(xstart-(onefifth),
ystart+k,
ptr);
}
ajGraphicsDrawposTextAtlineJustify(
xstart-(3*onefifth),
ystart+(flen2/(float)2.0),
xstart-(3*onefifth),ystart+flen2,
ajStrGetPtr(ajSeqsetGetseqNameS(seqset, j)),0.5);
}
ystart += flen2+(float)gap;
}
}
embWordFreeTable(&seq1MatchTable);
seq1MatchTable = NULL;
xstart += flen1+(float)gap;
ystart = 0.0;
}
ajGraphicsDrawposTextAtstart(total+onefifth,total-(onefifth),
"No. Length Lines Points Sequence");
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
seq1 = ajSeqsetGetseqSeq(seqset, i);
ajFmtPrintS(&sajb,"%3u %6d %5d %6d %s",i+1,
ajSeqGetLen(seq1),lines[i],
pts[i],ajSeqGetNameC(seq1));
ajGraphicsDrawposTextAtstart(total+onefifth,total-(onefifth*(i+2)),
ajStrGetPtr(sajb));
}
if(dumpfeat && seq1out)
{
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
ajFeattableWrite(seq1out, tabptr[i]);
ajFeattableDel(&tabptr[i]);
}
}
ajGraphicsClose();
ajGraphxyDel(&graph);
ajStrDel(&sajb);
AJFREE(lines);
AJFREE(pts);
AJFREE(tabptr);
ajSeqsetDel(&seqset);
ajFeattabOutDel(&seq1out);;
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqall queryseqs;
AjPSeqset targetseqs;
AjPSeq queryseq;
const AjPSeq targetseq;
AjPStr queryaln = 0;
AjPStr targetaln = 0;
AjPFile errorf;
AjBool show = ajFalse;
const char *queryseqc;
const char *targetseqc;
AjPMatrixf matrix;
AjPSeqCvt cvt = 0;
float **sub;
ajint *compass = NULL;
float *path = NULL;
float gapopen;
float gapextend;
float score;
float minscore;
ajuint j, k;
ajint querystart = 0;
ajint targetstart = 0;
ajint queryend = 0;
ajint targetend = 0;
ajint width = 0;
AjPTable kmers = 0;
ajint wordlen = 6;
ajint oldmax = 0;
ajint newmax = 0;
ajuint ntargetseqs;
ajuint nkmers;
AjPAlign align = NULL;
EmbPWordMatch maxmatch; /* match with maximum score */
/* Cursors for the current sequence being scanned,
** i.e., until which location it was scanned.
** Separate cursor/location entries for each sequence in the seqset.
*/
ajuint* lastlocation;
EmbPWordRK* wordsw = NULL;
AjPList* matchlist = NULL;
embInit("supermatcher", argc, argv);
matrix = ajAcdGetMatrixf("datafile");
queryseqs = ajAcdGetSeqall("asequence");
targetseqs= ajAcdGetSeqset("bsequence");
gapopen = ajAcdGetFloat("gapopen");
gapextend = ajAcdGetFloat("gapextend");
wordlen = ajAcdGetInt("wordlen");
align = ajAcdGetAlign("outfile");
errorf = ajAcdGetOutfile("errorfile");
width = ajAcdGetInt("width"); /* width for banded Smith-Waterman */
minscore = ajAcdGetFloat("minscore");
gapopen = ajRoundFloat(gapopen, 8);
gapextend = ajRoundFloat(gapextend, 8);
sub = ajMatrixfGetMatrix(matrix);
cvt = ajMatrixfGetCvt(matrix);
embWordLength(wordlen);
/* seqset sequence is the reference sequence for SAM format */
ajAlignSetRefSeqIndx(align, 1);
ajSeqsetTrim(targetseqs);
ntargetseqs = ajSeqsetGetSize(targetseqs);
AJCNEW0(matchlist, ntargetseqs);
/* get tables of words */
for(k=0;k<ntargetseqs;k++)
{
targetseq = ajSeqsetGetseqSeq(targetseqs, k);
embWordGetTable(&kmers, targetseq);
ajDebug("Number of distinct kmers found so far: %d\n",
ajTableGetLength(kmers));
}
AJCNEW0(lastlocation, ntargetseqs);
if(ajTableGetLength(kmers)<1)
ajErr("no kmers found");
nkmers = embWordRabinKarpInit(kmers, &wordsw, wordlen, targetseqs);
while(ajSeqallNext(queryseqs,&queryseq))
{
ajSeqTrim(queryseq);
queryaln = ajStrNewRes(1+ajSeqGetLen(queryseq));
ajDebug("Read '%S'\n", ajSeqGetNameS(queryseq));
for(k=0;k<ntargetseqs;k++)
{
lastlocation[k]=0;
matchlist[k] = ajListstrNew();
}
embWordRabinKarpSearch(ajSeqGetSeqS(queryseq), targetseqs,
(const EmbPWordRK*)wordsw, wordlen, nkmers,
matchlist, lastlocation, ajFalse);
for(k=0;k<ajSeqsetGetSize(targetseqs);k++)
{
targetseq = ajSeqsetGetseqSeq(targetseqs, k);
ajDebug("Processing '%S'\n", ajSeqGetNameS(targetseq));
if(ajListGetLength(matchlist[k])==0)
{
ajFmtPrintF(errorf,
"No wordmatch start points for "
"%s vs %s. No alignment\n",
ajSeqGetNameC(queryseq),ajSeqGetNameC(targetseq));
embWordMatchListDelete(&matchlist[k]);
continue;
}
/* only the maximum match is used as seed
* (if there is more than one location with the maximum match
* only the first one is used)
* TODO: we should add a new option to make above limit optional
*/
maxmatch = embWordMatchFirstMax(matchlist[k]);
supermatcher_findendpoints(maxmatch,targetseq, queryseq,
&targetstart, &querystart,
&targetend, &queryend);
targetaln=ajStrNewRes(1+ajSeqGetLen(targetseq));
queryseqc = ajSeqGetSeqC(queryseq);
targetseqc = ajSeqGetSeqC(targetseq);
ajStrAssignC(&queryaln,"");
ajStrAssignC(&targetaln,"");
ajDebug("++ %S v %S start:%d %d end:%d %d\n",
ajSeqGetNameS(targetseq), ajSeqGetNameS(queryseq),
targetstart, querystart, targetend, queryend);
newmax = (targetend-targetstart+2)*width;
if(newmax > oldmax)
{
AJCRESIZE0(path,oldmax,newmax);
AJCRESIZE0(compass,oldmax,newmax);
oldmax=newmax;
ajDebug("++ memory re/allocation for path/compass arrays"
" to size: %d\n", newmax);
}
else
{
AJCSET0(path,newmax);
AJCSET0(compass,newmax);
}
ajDebug("Calling embAlignPathCalcSWFast "
"%d..%d [%d/%d] %d..%d [%d/%d] width:%d\n",
querystart, queryend, (queryend - querystart + 1),
ajSeqGetLen(queryseq),
targetstart, targetend, (targetend - targetstart + 1),
ajSeqGetLen(targetseq),
width);
score = embAlignPathCalcSWFast(&targetseqc[targetstart],
&queryseqc[querystart],
targetend-targetstart+1,
queryend-querystart+1,
0,width,
gapopen,gapextend,
path,sub,cvt,
compass,show);
if(score>minscore)
{
embAlignWalkSWMatrixFast(path,compass,gapopen,gapextend,
targetseq,queryseq,
&targetaln,&queryaln,
targetend-targetstart+1,
queryend-querystart+1,
0,width,
&targetstart,&querystart);
if(!ajAlignFormatShowsSequences(align))
{
ajAlignDefineCC(align, ajStrGetPtr(targetaln),
ajStrGetPtr(queryaln),
ajSeqGetNameC(targetseq),
ajSeqGetNameC(queryseq));
ajAlignSetScoreR(align, score);
}
else
{
ajDebug(" queryaln:%S \ntargetaln:%S\n",
queryaln,targetaln);
embAlignReportLocal(align,
queryseq, targetseq,
queryaln, targetaln,
querystart, targetstart,
gapopen, gapextend,
score, matrix,
1 + ajSeqGetOffset(queryseq),
1 + ajSeqGetOffset(targetseq)
);
}
ajAlignWrite(align);
ajAlignReset(align);
}
ajStrDel(&targetaln);
embWordMatchListDelete(&matchlist[k]);
}
ajStrDel(&queryaln);
}
for(k=0;k<nkmers;k++)
{
AJFREE(wordsw[k]->seqindxs);
AJFREE(wordsw[k]->nSeqMatches);
for(j=0;j<wordsw[k]->nseqs;j++)
AJFREE(wordsw[k]->locs[j]);
AJFREE(wordsw[k]->nnseqlocs);
AJFREE(wordsw[k]->locs);
AJFREE(wordsw[k]);
}
embWordFreeTable(&kmers);
if(!ajAlignFormatShowsSequences(align))
ajMatrixfDel(&matrix);
AJFREE(path);
AJFREE(compass);
AJFREE(kmers);
AJFREE(wordsw);
AJFREE(matchlist);
AJFREE(lastlocation);
ajAlignClose(align);
ajAlignDel(&align);
ajSeqallDel(&queryseqs);
ajSeqDel(&queryseq);
ajSeqsetDel(&targetseqs);
ajFileClose(&errorf);
embExit();
return 0;
}
int main(int argc, char **argv)
{
AjPSeqset seqset;
AjPSeqall seqall;
AjPSeq queryseq;
const AjPSeq targetseq;
ajint wordlen;
AjPTable wordsTable = NULL;
AjPList* matchlist = NULL;
AjPFile logfile;
AjPFeattable* seqsetftables = NULL;
AjPFeattable seqallseqftable = NULL;
AjPFeattabOut ftoutforseqsetseq = NULL;
AjPFeattabOut ftoutforseqallseq = NULL;
AjPAlign align = NULL;
AjIList iter = NULL;
ajint targetstart;
ajint querystart;
ajint len;
ajuint i, j;
ajulong nAllMatches = 0;
ajulong sumAllScore = 0;
AjBool dumpAlign = ajTrue;
AjBool dumpFeature = ajTrue;
AjBool checkmode = ajFalse;
EmbPWordRK* wordsw = NULL;
ajuint npatterns = 0;
ajuint seqsetsize;
ajuint nmatches;
ajuint* nmatchesseqset;
ajuint* lastlocation; /* Cursors for Rabin-Karp search. */
/* Shows until what point the query sequence was
* scanned for a pattern sequences in the seqset.
*/
char* paddedheader = NULL;
const char* header;
AjPStr padding;
header = "Pattern %S #pat-sequences #all-matches avg-match-length\n";
padding = ajStrNew();
embInit("wordmatch", argc, argv);
wordlen = ajAcdGetInt("wordsize");
seqset = ajAcdGetSeqset("asequence");
seqall = ajAcdGetSeqall("bsequence");
logfile = ajAcdGetOutfile("logfile");
dumpAlign = ajAcdGetToggle("dumpalign");
dumpFeature = ajAcdGetToggle("dumpfeat");
if(dumpAlign)
{
align = ajAcdGetAlign("outfile");
ajAlignSetExternal(align, ajTrue);
}
seqsetsize = ajSeqsetGetSize(seqset);
ajSeqsetTrim(seqset);
AJCNEW0(matchlist, seqsetsize);
AJCNEW0(seqsetftables, seqsetsize);
AJCNEW0(nmatchesseqset, seqsetsize);
if (dumpFeature)
{
ftoutforseqsetseq = ajAcdGetFeatout("aoutfeat");
ftoutforseqallseq = ajAcdGetFeatout("boutfeat");
}
checkmode = !dumpFeature && !dumpAlign;
embWordLength(wordlen);
ajFmtPrintF(logfile, "Small sequence/file for constructing"
" target patterns: %S\n", ajSeqsetGetUsa(seqset));
ajFmtPrintF(logfile, "Large sequence/file to be scanned"
" for patterns: %S\n", ajSeqallGetUsa(seqall));
ajFmtPrintF(logfile, "Number of sequences in the patterns file: %u\n",
seqsetsize);
ajFmtPrintF(logfile, "Pattern/word length: %u\n", wordlen);
for(i=0;i<seqsetsize;i++)
{
targetseq = ajSeqsetGetseqSeq(seqset, i);
embWordGetTable(&wordsTable, targetseq);
}
AJCNEW0(lastlocation, seqsetsize);
if(ajTableGetLength(wordsTable)>0)
{
npatterns = embWordRabinKarpInit(wordsTable,
&wordsw, wordlen, seqset);
ajFmtPrintF(logfile, "Number of patterns/words found: %u\n", npatterns);
while(ajSeqallNext(seqall,&queryseq))
{
for(i=0;i<seqsetsize;i++)
{
lastlocation[i]=0;
if (!checkmode)
matchlist[i] = ajListstrNew();
}
nmatches = embWordRabinKarpSearch(
ajSeqGetSeqS(queryseq), seqset,
(EmbPWordRK const *)wordsw, wordlen, npatterns,
matchlist, lastlocation, checkmode);
nAllMatches += nmatches;
if (checkmode)
continue;
for(i=0;i<seqsetsize;i++)
{
if(ajListGetLength(matchlist[i])>0)
{
iter = ajListIterNewread(matchlist[i]) ;
while(embWordMatchIter(iter, &targetstart, &querystart, &len,
&targetseq))
{
if(dumpAlign)
{
ajAlignDefineSS(align, targetseq, queryseq);
ajAlignSetScoreI(align, len);
/* ungapped alignment means same length
* for both sequences
*/
ajAlignSetSubRange(align, targetstart, 1, len,
ajSeqIsReversed(targetseq),
ajSeqGetLen(targetseq),
querystart, 1, len,
ajSeqIsReversed(queryseq),
ajSeqGetLen(queryseq));
}
}
if(dumpAlign)
{
ajAlignWrite(align);
ajAlignReset(align);
}
if(ajListGetLength(matchlist[i])>0 && dumpFeature)
{
embWordMatchListConvToFeat(matchlist[i],
&seqsetftables[i],
&seqallseqftable,
targetseq, queryseq);
ajFeattableWrite(ftoutforseqallseq, seqallseqftable);
ajFeattableDel(&seqallseqftable);
}
ajListIterDel(&iter);
}
embWordMatchListDelete(&matchlist[i]);
}
}
/* search completed, now report statistics */
for(i=0;i<npatterns;i++)
{
sumAllScore += wordsw[i]->lenMatches;
for(j=0;j<wordsw[i]->nseqs;j++)
nmatchesseqset[wordsw[i]->seqindxs[j]] +=
wordsw[i]->nSeqMatches[j];
}
ajFmtPrintF(logfile, "Number of sequences in the file scanned "
"for patterns: %u\n", ajSeqallGetCount(seqall));
ajFmtPrintF(logfile, "Number of all matches: %Lu"
" (wordmatch finds exact matches only)\n", nAllMatches);
if(nAllMatches>0)
{
ajFmtPrintF(logfile, "Sum of match lengths: %Lu\n", sumAllScore);
ajFmtPrintF(logfile, "Average match length: %.2f\n",
sumAllScore*1.0/nAllMatches);
ajFmtPrintF(logfile, "\nDistribution of the matches among pattern"
" sequences:\n");
ajFmtPrintF(logfile, "-----------------------------------------"
"-----------\n");
for(i=0;i<ajSeqsetGetSize(seqset);i++)
{
if (nmatchesseqset[i]>0)
ajFmtPrintF(logfile, "%-42s: %8u\n",
ajSeqGetNameC(ajSeqsetGetseqSeq(seqset, i)),
nmatchesseqset[i]);
ajFeattableWrite(ftoutforseqsetseq, seqsetftables[i]);
ajFeattableDel(&seqsetftables[i]);
}
ajFmtPrintF(logfile, "\nPattern statistics:\n");
ajFmtPrintF(logfile, "-------------------\n");
if(wordlen>7)
ajStrAppendCountK(&padding, ' ', wordlen-7);
paddedheader = ajFmtString(header,padding);
ajFmtPrintF(logfile, paddedheader);
for(i=0;i<npatterns;i++)
if (wordsw[i]->nMatches>0)
ajFmtPrintF(logfile, "%-7s: %12u %12u %17.2f\n",
wordsw[i]->word->fword, wordsw[i]->nseqs,
wordsw[i]->nMatches,
wordsw[i]->lenMatches*1.0/wordsw[i]->nMatches);
}
}
for(i=0;i<npatterns;i++)
{
for(j=0;j<wordsw[i]->nseqs;j++)
AJFREE(wordsw[i]->locs[j]);
AJFREE(wordsw[i]->locs);
AJFREE(wordsw[i]->seqindxs);
AJFREE(wordsw[i]->nnseqlocs);
AJFREE(wordsw[i]->nSeqMatches);
AJFREE(wordsw[i]);
}
embWordFreeTable(&wordsTable);
AJFREE(wordsw);
AJFREE(matchlist);
AJFREE(lastlocation);
AJFREE(nmatchesseqset);
AJFREE(seqsetftables);
if(dumpAlign)
{
ajAlignClose(align);
ajAlignDel(&align);
}
if(dumpFeature)
{
ajFeattabOutDel(&ftoutforseqsetseq);
ajFeattabOutDel(&ftoutforseqallseq);
}
ajFileClose(&logfile);
ajSeqallDel(&seqall);
ajSeqsetDel(&seqset);
ajSeqDel(&queryseq);
ajStrDel(&padding);
AJFREE(paddedheader);
embExit();
return 0;
}
