static GtUchar gt_mmsearch_accessquery(const GtQueryrepresentation *queryrep,
GtUword pos)
{
GtUword abspos, cc;
gt_assert(queryrep != NULL);
gt_assert(pos < queryrep->seqlen);
abspos = queryrep->startpos + (queryrep->readmode == GT_READMODE_FORWARD
? pos
: GT_REVERSEPOS(queryrep->seqlen,pos));
if (queryrep->sequence != NULL)
{
cc = queryrep->sequence[abspos];
} else
{
gt_assert(queryrep->encseq != NULL);
cc = gt_encseq_get_encoded_char(queryrep->encseq,abspos,
GT_READMODE_FORWARD);
}
if (GT_ISDIRCOMPLEMENT(queryrep->readmode))
{
if (ISSPECIAL(cc))
{
return cc;
}
return GT_COMPLEMENTBASE(cc);
} else
{
return cc;
}
}
static GtUchar gt_mmsearch_accessquery(const GtQueryrepresentation *queryrep,
GtUword pos)
{
GtUword abspos;
gt_assert(queryrep != NULL);
gt_assert(pos < queryrep->seqlen);
abspos = queryrep->startpos + (queryrep->readmode == GT_READMODE_FORWARD
? pos
: GT_REVERSEPOS(queryrep->seqlen,pos));
if (queryrep->sequence != NULL)
{
gt_assert(!GT_ISDIRCOMPLEMENT(queryrep->readmode)); /* not implemented */
return queryrep->sequence[abspos];
} else
{
gt_assert(queryrep->encseq != NULL);
return gt_encseq_get_encoded_char(queryrep->encseq,abspos,
GT_READMODE_FORWARD);
}
}
void gt_edistmyersbitvectorAPM(Myersonlineresources *mor,
const GtUchar *pattern,
GtUword patternlength,
GtUword maxdistance)
{
GtUword Pv = ~0UL,
Mv = 0UL,
Eq,
Xv,
Xh,
Ph,
Mh,
score;
const GtUword Ebit = 1UL << (patternlength-1);
GtUchar cc;
GtUword pos;
const GtReadmode readmode = GT_READMODE_REVERSE;
GtIdxMatch match;
gt_initeqsvectorrev(mor->eqsvectorrev,
(GtUword) mor->alphasize,
pattern,patternlength);
score = patternlength;
gt_encseq_reader_reinit_with_readmode(mor->esr, mor->encseq, readmode, 0);
match.dbabsolute = NULL;
match.dbsubstring = NULL;
match.querystartpos = 0;
match.querylen = patternlength;
match.alignment = NULL;
for (pos = 0; pos < mor->totallength; pos++)
{
cc = gt_encseq_reader_next_encoded_char(mor->esr);
if (cc == (GtUchar) SEPARATOR)
{
Pv = ~0UL;
Mv = 0UL;
score = patternlength;
} else
{
if (cc == (GtUchar) WILDCARD)
{
Eq = 0;
} else
{
Eq = mor->eqsvectorrev[(GtUword) cc]; /* 6 */
}
Xv = Eq | Mv; /* 7 */
Xh = (((Eq & Pv) + Pv) ^ Pv) | Eq; /* 8 */
Ph = Mv | ~ (Xh | Pv); /* 9 */
Mh = Pv & Xh; /* 10 */
if (Ph & Ebit)
{
score++;
} else
{
if (Mh & Ebit)
{
gt_assert(score > 0);
score--;
}
}
Ph <<= 1; /* 15 */
Pv = (Mh << 1) | ~ (Xv | Ph); /* 17 */
Mv = Ph & Xv; /* 18 */
if (score <= maxdistance)
{
GtUword dbstartpos = GT_REVERSEPOS(mor->totallength,pos);
Definedunsignedlong matchlength;
if (maxdistance > 0)
{
matchlength = gt_forwardprefixmatch(mor->encseq,
mor->alphasize,
dbstartpos,
mor->nowildcards,
mor->eqsvector,
pattern,
patternlength,
maxdistance);
} else
{
matchlength.defined = true;
matchlength.valueunsignedlong = patternlength;
}
gt_assert(matchlength.defined || mor->nowildcards);
if (matchlength.defined)
{
match.dbstartpos = dbstartpos;
match.dblen = (GtUword) matchlength.valueunsignedlong;
match.distance = score;
mor->processmatch(mor->processmatchinfo,&match);
}
}
}
}
}
static void seededmatch2alignment(const GtQuerymatch *querymatch,
GtUword querystartabsolute,
const GtEncseq *encseq,
GtQuerymatchoutoptions *querymatchoutoptions)
{
GtUword ulen, vlen, rightdistance = 0, leftdistance = 0,
offset1 = querymatchoutoptions->seedpos1 +
querymatchoutoptions->seedlen,
offset2 = querymatchoutoptions->seedpos2 +
querymatchoutoptions->seedlen;
GtAlignment *alignment = gt_alignment_new();
gt_assert(querymatch->dbstart + querymatch->dblen >= offset1);
ulen = querymatch->dbstart + querymatch->dblen - offset1;
gt_assert(querystartabsolute + querymatch->querylen >= offset2);
vlen = querystartabsolute + querymatch->querylen - offset2;
if (ulen > 0 && vlen > 0)
{
Polished_point best_polished_point = {0,0,0};
rightdistance = align_front_prune_edist(true,
&best_polished_point,
querymatchoutoptions->front_trace,
encseq,
querymatchoutoptions->ggemi,
offset1,
ulen,
offset2,
vlen);
front_trace2alignments(alignment,
querymatchoutoptions->front_trace,
&best_polished_point,
ulen,vlen);
front_trace_reset(querymatchoutoptions->front_trace,ulen+vlen);
}
gt_assert(rightdistance <= querymatch->edist);
if (querymatchoutoptions->totallength == GT_UWORD_MAX)
{
querymatchoutoptions->totallength = gt_encseq_total_length(encseq);
}
if (querymatchoutoptions->seedpos1 > querymatch->dbstart &&
querymatchoutoptions->seedpos2 > querystartabsolute)
{
Polished_point best_polished_point = {0,0,0};
ulen = querymatchoutoptions->seedpos1 - querymatch->dbstart;
vlen = querymatchoutoptions->seedpos2 - querystartabsolute;
leftdistance = align_front_prune_edist(false,
&best_polished_point,
querymatchoutoptions->front_trace,
encseq,
querymatchoutoptions->ggemi,
GT_REVERSEPOS(querymatchoutoptions->totallength,
querymatchoutoptions->seedpos1-1),
ulen,
GT_REVERSEPOS(querymatchoutoptions->totallength,
querymatchoutoptions->seedpos2-1),
vlen);
front_trace_reset(querymatchoutoptions->front_trace,ulen+vlen);
}
if (leftdistance + rightdistance > querymatch->edist)
{
fprintf(stderr,"leftdistance + rightdistance = " GT_WU " + " GT_WU " > "
GT_WU " = querymatch->edist\n",
leftdistance,rightdistance,querymatch->edist);
}
gt_assert(leftdistance + rightdistance <= querymatch->edist);
gt_alignment_delete(alignment);
}
