static void init_missed_line(args_t *args)
{
int i;
for (i=0; i<bcf_hdr_nsamples(args->aux.hdr); i++)
{
args->aux.gts[i*2] = bcf_gt_missing;
args->aux.gts[i*2+1] = bcf_int32_vector_end;
}
args->missed_line = bcf_init1();
bcf_update_genotypes(args->aux.hdr, args->missed_line, args->aux.gts, 2*bcf_hdr_nsamples(args->aux.hdr));
bcf_float_set_missing(args->missed_line->qual);
}
/**
* @brief Construct individual field objects and create a mapping from field logical (header) index to
* physical locations in the m_*_fields vectors.
*/
void VariantBuilderIndividualRegion::build_lookup_tables() {
// Must copy float missing/eov values into local variables; using bcf_float_missing and bcf_float_vector_end
// values directly doesn't work
auto float_missing = 0.0f; bcf_float_set_missing(float_missing);
auto float_vector_end = 0.0f; bcf_float_set_vector_end(float_vector_end);
const auto num_samples = m_header.n_samples();
// Create an entry for all possible BCF_DT_ID indices in the VariantHeader
for ( auto i = 0; i < m_header.m_header->n[BCF_DT_ID]; ++i ) {
if ( m_header.has_individual_field(i) ) {
const auto field_type = m_header.individual_field_type(i);
switch ( field_type ) {
case BCF_HT_INT :
m_int_fields.emplace_back(num_samples, i, field_type, bcf_int32_missing, bcf_int32_vector_end, int_field_short_value_threshold);
m_field_lookup_table[i] = m_int_fields.size() - 1;
break;
case BCF_HT_REAL :
m_float_fields.emplace_back(num_samples, i, field_type, float_missing, float_vector_end, float_field_short_value_threshold);
m_field_lookup_table[i] = m_float_fields.size() - 1;
break;
case BCF_HT_STR :
// GT is a string field in the header, although we encode it as an int field
if ( i == m_gt_field_index ) {
// Note that the missing value for genotypes is 0, not bcf_int32_missing
m_int_fields.emplace_back(num_samples, i, field_type, 0, bcf_int32_vector_end, int_field_short_value_threshold);
m_field_lookup_table[i] = m_int_fields.size() - 1;
}
else {
// TODO: should we use bcf_str_missing here? htslib generally uses '.' instead of it...
m_string_fields.emplace_back(num_samples, i, field_type, '.', bcf_str_vector_end, string_field_short_value_threshold);
m_field_lookup_table[i] = m_string_fields.size() - 1;
}
break;
default:
throw logic_error(string{"Found format field in header with unsupported type: "} + to_string(field_type));
}
}
else {
m_field_lookup_table[i] = -1;
}
}
}
void VariantBuilder::build_from_scratch(const std::shared_ptr<bcf1_t>& new_variant_body) const {
// Missing rid/pos are errors that will be caught in post-build validation (if validation is turned on)
new_variant_body->rid = m_contig.is_set() ? m_contig.field_value() : missing_values::int32;
new_variant_body->pos = m_start_pos.is_set() ? m_start_pos.field_value() : missing_values::int32;
if ( m_qual.is_set() ) {
new_variant_body->qual = m_qual.field_value();
}
else {
bcf_float_set_missing(new_variant_body->qual);
}
// Set rlen to the reference block size if alignment stop was set, otherwise set it to the ref allele length.
// If m_start_pos is not set, we'll get an error in the post-validation stage.
new_variant_body->rlen = m_stop_pos.is_set() ? (m_stop_pos.field_value() - m_start_pos.field_value() + 1) :
int32_t(m_shared_region.ref_allele_length());
new_variant_body->n_allele = 1 + m_shared_region.num_alt_alleles();
// Shared region (always encoded, since at a minimum the ref allele will be present)
new_variant_body->n_info = m_shared_region.num_present_info_fields();
auto shared_buffer = utils::initialize_htslib_buffer(m_shared_region.estimate_total_size());
m_shared_region.encode_into(&shared_buffer);
new_variant_body->shared = shared_buffer;
// Individual region (conditionally encoded)
new_variant_body->n_sample = m_header.n_samples();
new_variant_body->n_fmt = m_individual_region.num_present_fields();
if ( m_individual_region.num_present_fields() > 0 ) {
auto indiv_buffer = utils::initialize_htslib_buffer(m_individual_region.estimate_total_size());
m_individual_region.encode_into(&indiv_buffer);
new_variant_body->indiv = indiv_buffer;
}
else {
new_variant_body->indiv = {0, 0, 0};
}
}
VariantBuilderMultiSampleVector<float> VariantBuilder::get_float_multi_sample_vector(const uint32_t num_samples, const uint32_t max_values_per_sample) const {
auto float_missing = 0.0f; bcf_float_set_missing(float_missing);
auto float_vector_end = 0.0f; bcf_float_set_vector_end(float_vector_end);
return VariantBuilderMultiSampleVector<float>{num_samples, max_values_per_sample, float_missing, float_vector_end};
}