def __getitem__(self, idx):
"""Returns (pybedtools.Interval, labels)
"""
row = self.df.iloc[idx]
interval = pybedtools.create_interval_from_list(
[to_scalar(x) for x in row.iloc[:self.bed_columns]])
if self.ignore_targets or self.n_tasks == 0:
labels = {}
else:
labels = row.iloc[self.bed_columns:].values.astype(
self.label_dtype)
return interval, labels
def __getitem__(self, idx):
if self.fasta_extractor is None:
self.fasta_extractor = FastaStringExtractor(
self.fasta_file, use_strand=True, force_upper=self.force_upper)
if self.vcf is None:
self.vcf = MultiSampleVCF(self.vcf_file)
if self.vcf_extractor is None:
self.vcf_extractor = VariantSeqExtractor(self.fasta_file)
entry = self.bed.iloc[idx]
entry_id = entry["id"]
entry_chr = entry["chr"]
entry_pos = entry["pos"]
entry_strand = entry["strand"]
ref_exons = []
var_exons = []
exon_pos_strings = []
exon_var_strings = []
for exon in entry_pos:
# We get the interval
interval = pybedtools.Interval(to_scalar(entry_chr),
to_scalar(exon[0]),
to_scalar(exon[1]),
strand=to_scalar(entry_strand))
exon_pos_strings.append("%s-%s" % (str(exon[0]), str(exon[1])))
# We get the reference sequence
ref_seq = self.fasta_extractor.extract(interval)
# We get the variants, insert them and also save them as metadata
variants = list(self.vcf.fetch_variants(interval))
if len(variants) == 0:
ref_exons.append(ref_seq)
var_exons.append(ref_seq)
else:
var_seq = self.vcf_extractor.extract(interval,
variants=variants,
anchor=0,
fixed_len=False)
var_string = ";".join([str(var) for var in variants])
ref_exons.append(ref_seq)
var_exons.append(var_seq)
exon_var_strings.append(var_string)
# Combine
if entry_strand == "-":
ref_exons.reverse()
var_exons.reverse()
ref_seq = "".join(ref_exons)
var_seq = "".join(var_exons)
pos_string = ";".join(exon_pos_strings)
var_string = ";".join(exon_var_strings)
return {
"inputs": {
"ref_seq": ref_seq,
"alt_seq": var_seq,
},
"metadata": {
"id": entry_id,
"chr": entry_chr,
"exon_positions": pos_string,
"strand": entry_strand,
"variants": var_string
}
}