VariantContig::VariantContig(
RawVariant const& var,
Fasta& ref,
int flank,
std::string const& seqname
)
{
uint64_t seqlen = ref.seqlen(seqname);
uint64_t preflank_len = var.pos <= flank ? var.pos - 1 : flank;
_start = std::max(1ul, var.pos - preflank_len);
_stop = std::min(var.pos + var.ref.size() - 1 + flank, seqlen);
uint64_t postflank_start = var.pos + var.ref.size();
uint64_t postflank_len = _stop - postflank_start + 1;
// build sequence
if (preflank_len)
_sequence = ref.sequence(seqname, _start, preflank_len); // left flank
_sequence += var.alt;
if (postflank_start <= seqlen && postflank_len)
_sequence += ref.sequence(seqname, postflank_start, postflank_len); // right flank
// build cigar
if (preflank_len)
_cigar.push_back(preflank_len, MATCH);
if (var.ref.size() > var.alt.size()) {
_cigar.push_back(var.alt.size(), MATCH);
_cigar.push_back(var.ref.size() - var.alt.size(), DEL);
} else if (var.ref.size() < var.alt.size()) {
_cigar.push_back(var.ref.size(), MATCH);
_cigar.push_back(var.alt.size() - var.ref.size(), INS);
} else {
_cigar.push_back(var.alt.size(), MATCH);
}
if (postflank_len)
_cigar.push_back(postflank_len, MATCH);
}
void align_against_collection(string &read, Fasta &rep, int forbidden_rep_id,
bool reverse_ref, bool reverse_both, bool local,
AlignBox *box, Cost segment_cost)
{
int best_score = MINUS_INF ;
box->ref_nb = MINUS_INF ;
int best_best_i = (int) string::npos ;
int best_best_j = (int) string::npos ;
int best_first_i = (int) string::npos ;
int best_first_j = (int) string::npos ;
vector<pair<int, int> > score_r;
DynProg::DynProgMode dpMode = DynProg::LocalEndWithSomeDeletions;
if (local==true) dpMode = DynProg::Local;
// With reverse_ref, the read is reversed to prevent calling revcomp on each reference sequence
string sequence_or_rc = revcomp(read, reverse_ref);
for (int r = 0 ; r < rep.size() ; r++)
{
if (r == forbidden_rep_id)
continue;
DynProg dp = DynProg(sequence_or_rc, rep.sequence(r),
dpMode, // DynProg::SemiGlobalTrans,
segment_cost, // DNA
reverse_both, reverse_both,
rep.read(r).marked_pos);
bool onlyBottomTriangle = !local ;
int score = dp.compute(onlyBottomTriangle, BOTTOM_TRIANGLE_SHIFT);
if (local==true){
dp.backtrack();
}
if (score > best_score)
{
best_score = score ;
best_best_i = dp.best_i ;
best_best_j = dp.best_j ;
best_first_i = dp.first_i ;
best_first_j = dp.first_j ;
box->ref_nb = r ;
box->ref_label = rep.label(r) ;
if (!local)
dp.backtrack();
box->marked_pos = dp.marked_pos_i ;
}
score_r.push_back(make_pair(score, r));
// #define DEBUG_SEGMENT
#ifdef DEBUG_SEGMENT
cout << rep.label(r) << " " << score << " " << dp.best_i << endl ;
#endif
}
sort(score_r.begin(),score_r.end(),comp_pair);
box->ref = rep.sequence(box->ref_nb);
box->del_right = reverse_both ? best_best_j : box->ref.size() - best_best_j - 1;
box->del_left = best_first_j;
box->start = best_first_i;
box->score = score_r;
#ifdef DEBUG_SEGMENT
cout << "best: " << box->ref_label << " " << best_score ;
cout << "del/del2/begin:" << (box->del_right) << "/" << (box->del_left) << "/" << (box->start) << endl;
cout << endl;
#endif
if (reverse_ref)
// Why -1 here and +1 in dynprog.cpp /// best_i = m - best_i + 1 ;
best_best_i = read.length() - best_best_i - 1 ;
box->end = best_best_i ;
}