/**
* Returns true if chrom:start1-end1 overlaps with a region in the file.
*/
bool OrderedBCFOverlapMatcher::overlaps_with(std::string& chrom, int32_t start1, int32_t end1)
{
bool overlaps = false;
//moves to new chromosome
if (current_interval.seq!=chrom)
{
buffer.clear();
current_interval.set(chrom);
odr->jump_to_interval(current_interval);
std::cerr << "Jumped to chromosome " << chrom << "\n";
while (odr->read(v))
{
if (bcf_get_end_pos1(v)<start1) continue;
overlaps = overlaps || (bcf_get_pos1(v)<=end1);
buffer.push_back(v);
if (bcf_get_pos1(v)>end1) break;
v = bcf_init();
}
}
else
{
//scythe preceding bed records
std::list<bcf1_t*>::iterator i = buffer.begin();
while (i!=buffer.end())
{
if (bcf_get_end_pos1(*i)<start1)
{
bcf_destroy(*i);
i = buffer.erase(i);
continue;
}
overlaps = (bcf_get_pos1(*i)<=end1);
break;
}
if (!overlaps)
{
if (buffer.empty())
{
while (odr->read(v))
{
if (bcf_get_end_pos1(*i)<start1) continue;
overlaps = overlaps || (bcf_get_pos1(*i)<=end1);
buffer.push_back(v);
if (bcf_get_pos1(v)>end1) break;
v = bcf_init();
}
}
}
}
return overlaps;
};
/**
* Constructor.
* @v - VCF record.
*/
GenotypingRecord::GenotypingRecord(bcf_hdr_t *h, bcf1_t *v, int32_t vtype)
{
clear();
this->h = h;
this->v = v;
rid = bcf_get_rid(v);
pos1 = bcf_get_pos1(v);
this->vtype = vtype;
int32_t n_allele = bcf_get_n_allele(v);
if (vtype==VT_SNP && n_allele==2)
{
rid = bcf_get_rid(v);
beg1 = bcf_get_pos1(v);
end1 = beg1;
}
else if (vtype==VT_INDEL && bcf_get_n_allele(v)==2)
{
rid = bcf_get_rid(v);
char** alleles = bcf_get_allele(v);
dlen = strlen(alleles[1])-strlen(alleles[0]);
len = abs(dlen);
int32_t *flanks = NULL;
int32_t n = 0;
if (bcf_get_info_int32(h, v, "FLANKS", &flanks, &n)>0)
{
lend1 = flanks[0];
rbeg1 = flanks[1];
free(flanks);
}
else
{
lend1 = bcf_get_pos1(v) - 1;
rbeg1 = bcf_get_end_pos1(v) + 1;
}
int32_t *fuzzy_flanks = NULL;
n = 0;
if (bcf_get_info_int32(h, v, "FZ_FLANKS", &fuzzy_flanks, &n)>0)
{
fuzzy_lend1 = fuzzy_flanks[0];
fuzzy_rbeg1 = fuzzy_flanks[1];
free(fuzzy_flanks);
}
else
{
fuzzy_lend1 = bcf_get_pos1(v) - 1;
fuzzy_rbeg1 = bcf_get_end_pos1(v) + 1;
}
beg1 = std::min(lend1-2, fuzzy_lend1-2);
end1 = std::max(rbeg1+2, fuzzy_rbeg1+2);
//construct alleles
//get reference sequence
// char* ref_seq = NULL;
// int32_t ref_len = 0;
//// ref_seq = faidx_fetch_seq(fai, bcf_get_chrom(h,v), lend1+1-1, rbeg1-1-1, &ref_len);
//
// for (uint32_t i=0; i<n_allele; ++i)
// {
//
// }
// for ()
// {
// }
//
if (dlen>0)
{
indel.append(&alleles[1][1]);
}
else
{
indel.append(&alleles[0][1]);
}
}
else if (vtype==VT_VNTR)
{
rid = bcf_get_rid(v);
beg1 = bcf_get_pos1(v) - 1;
end1 = bcf_get_end_pos1(v) + 1;
char *motif = NULL;
int32_t n = 0;
if (bcf_get_info_string(h, v, "MOTIF", &motif, &n)>0)
{
this->motif.assign(motif);
free(motif);
}
}
}