static AjBool assemoutWriteBamAlignment(AjPSeqBamBgzf gzfile,
const AjPAssemRead r,
AjPSeqBam bam)
{
AjPSeqBamCore c;
AjPAssemTag tag;
unsigned char *dpos;
const char *s;
ajuint ilen;
ajuint slen;
ajuint i;
AjIList l = NULL;
/* optional fields */
ajint tagvalsize = 0;
const unsigned char* tagval = 0;
ajint intval =0;
/* processing cigar strings*/
char *t;
int op;
long x;
unsigned char bam_nt16_table[256] =
{
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
1, 2, 4, 8, 15,15,15,15, 15,15,15,15, 15, 0 /*=*/,15,15,
15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 15,15,15,15,
15, 1,14, 2, 13,15,15, 4, 11,15,15,12, 15, 3,15,15,
15,15, 5, 6, 8,15, 7, 9, 15,10,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15,
15,15,15,15, 15,15,15,15, 15,15,15,15, 15,15,15,15
};
/* bam_write1 for each alignment */
c = &bam->core;
ilen = ajStrGetLen(r->Seq);
c->tid = (int) r->Reference;
if(r->Flag & BAM_FREVERSE)
c->pos = r->y1-1;
else
c->pos = r->x1-1; /* BAM format is zero based;
-1 is translated to 0, meaning unmapped */
c->bin = 0;
c->qual = r->MapQ;
c->l_qname = 1 + ajStrGetLen(r->Name);
c->flag = r->Flag;
c->n_cigar = 0;
c->l_qseq = ilen;
c->mtid = (int) r->Rnext;
c->mpos = (int) r->Pnext-1;
c->isize = r->Tlen;
/* get cigar string length */
s = ajStrGetPtr(r->Cigar);
if (strcmp(s,"*")) /* '*' means unavailable */
{
for (; *s; ++s)
{
if ((isalpha((int)*s)) || (*s=='='))
++c->n_cigar;
else
if (!isdigit((int)*s))
ajWarn("invalid CIGAR character: %c\n", *s);
}
}
bam->data_len = c->n_cigar*4 + c->l_qname +
(ilen + 1)/2 + ilen;
/* allocation for optional tags are made as they are appended */
if(bam->data_len > bam->m_data)
{
AJCRESIZE0(bam->data,bam->m_data, bam->data_len);
bam->m_data = bam->data_len;
}
dpos = bam->data;
/* copy query name to bam->data */
memcpy(dpos, ajStrGetPtr(r->Name), c->l_qname);
dpos += c->l_qname;
/* copy cigar string to bam->data */
s = ajStrGetPtr(r->Cigar);
for (i = 0; i != c->n_cigar; ++i)
{
x = strtol(s, &t, 10);
op = toupper((int)*t);
if (op == 'M') op = BAM_CMATCH;
else if (op == 'I') op = BAM_CINS;
else if (op == 'D') op = BAM_CDEL;
else if (op == 'N') op = BAM_CREF_SKIP;
else if (op == 'S') op = BAM_CSOFT_CLIP;
else if (op == 'H') op = BAM_CHARD_CLIP;
else if (op == 'P') op = BAM_CPAD;
else if (op == '=') op = BAM_CEQUAL;
else if (op == 'X') op = BAM_CDIFF;
else ajWarn("invalid CIGAR operation: %c",op);
s = t + 1;
((ajuint*)dpos)[i] = x << BAM_CIGAR_SHIFT | op;
}
if (*s && c->n_cigar)
ajWarn("unmatched CIGAR operation: %c", *s);
c->bin = ajSeqBamReg2bin(c->pos, ajSeqBamCalend(c, MAJSEQBAMCIGAR(bam)));
dpos += c->n_cigar*4;
/* copy sequence string to bam->data */
s = ajStrGetPtr(r->Seq);
slen = (ilen+1)/2;
for (i = 0; i < slen; ++i)
dpos[i] = 0;
for (i = 0; i < ilen; ++i)
dpos[i/2] |= bam_nt16_table[(ajuint)s[i]] << 4*(1-i%2);
dpos += slen;
/* copy quality values to bam->data */
if(r->SeqQ && !ajStrMatchC(r->SeqQ, "*"))
{
s = ajStrGetPtr(r->SeqQ);
for(i=0;i<ilen;i++)
dpos[i]= s[i]-33;
}
else
for(i=0;i<ilen;i++)
dpos[i]= 0xff;
l = ajListIterNewread(r->Tags);
bam->l_aux=0;
while (!ajListIterDone(l))
{
tag = ajListIterGet(l);
/* TODO: array type 'B' and other types */
if(tag->type == 'i' || tag->type == 'I')
{
tagvalsize = 4;
ajStrToInt(tag->Comment, &intval);
tagval = (unsigned char*)&intval;
}
else if(tag->type =='s' || tag->type =='S')
{
tagvalsize = 2;
ajStrToInt(tag->Comment, &intval);
tagval = (unsigned char*)&intval;
}
else if(tag->type =='c' || tag->type =='C')
{
tagvalsize = 1;
ajStrToInt(tag->Comment, &intval);
tagval = (unsigned char*)&intval;
}
else if(tag->type =='A')
{
tagvalsize = 1;
tagval = (const unsigned char*)ajStrGetPtr(tag->Comment);
}
else if(tag->type =='Z')
{
tagvalsize = ajStrGetLen(tag->Comment)+1;
tagval = (const unsigned char*)ajStrGetPtr(tag->Comment);
}
else
{
ajWarn("tag type '%c' not yet supported",tag->type);
continue;
}
ajSeqBamAuxAppend(bam,
ajStrGetPtr(tag->Name),
tag->type,
tagvalsize,
tagval);
}
ajListIterDel(&l);
ajSeqBamWrite(gzfile, bam);
return ajTrue;
}
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;
}
