bcf1_t *process(bcf1_t *rec)
{
if ( args.prev_rid==-1 ) args.prev_rid = rec->rid;
if ( args.prev_rid!=rec->rid ) flush_viterbi(&args);
args.prev_rid = rec->rid;
args.set_observed_prob(rec);
return NULL;
}
void destroy(void)
{
flush_viterbi(&args);
fclose(args.fp);
free(args.gt_arr);
free(args.tprob);
free(args.sites);
free(args.eprob);
hmm_destroy(args.hmm);
}
static void vcfroh(args_t *args, bcf1_t *line)
{
// Are we done?
if ( !line )
{
flush_viterbi(args);
return;
}
args->ntot++;
// Skip unwanted lines
if ( line->rid == args->skip_rid ) return;
if ( line->n_allele==1 ) return; // no ALT allele
if ( line->n_allele!=2 ) return; // only biallelic sites
if ( args->snps_only && !bcf_is_snp(line) ) return;
// Initialize genetic map
int skip_rid = 0;
if ( args->prev_rid<0 )
{
args->prev_rid = line->rid;
args->prev_pos = line->pos;
skip_rid = load_genmap(args, line);
if ( !skip_rid && args->vi_training ) push_rid(args, line->rid);
}
// New chromosome?
if ( args->prev_rid!=line->rid )
{
skip_rid = load_genmap(args, line);
if ( args->vi_training )
{
if ( !skip_rid ) push_rid(args, line->rid);
}
else
{
flush_viterbi(args);
args->nsites = 0;
}
args->prev_rid = line->rid;
args->prev_pos = line->pos;
}
if ( skip_rid )
{
fprintf(pysamerr,"Skipping the sequence, no genmap for %s\n", bcf_seqname(args->hdr,line));
args->skip_rid = line->rid;
return;
}
if ( args->prev_pos > line->pos ) error("The file is not sorted?!\n");
args->prev_rid = line->rid;
args->prev_pos = line->pos;
// Ready for the new site
int m = args->msites;
hts_expand(uint32_t,args->nsites+1,args->msites,args->sites);
if ( args->msites!=m )
args->eprob = (double*) realloc(args->eprob,sizeof(double)*args->msites*2);
// Set likelihoods and alternate allele frequencies
double alt_freq, pdg[3];
if ( parse_line(args, line, &alt_freq, pdg)<0 ) return; // something went wrong
args->nused++;
// Calculate emission probabilities P(D|AZ) and P(D|HW)
double *eprob = &args->eprob[2*args->nsites];
eprob[STATE_AZ] = pdg[0]*(1-alt_freq) + pdg[2]*alt_freq;
eprob[STATE_HW] = pdg[0]*(1-alt_freq)*(1-alt_freq) + 2*pdg[1]*(1-alt_freq)*alt_freq + pdg[2]*alt_freq*alt_freq;
args->sites[args->nsites] = line->pos;
args->nsites++;
}
