def extract_recessive_disorder_candidates(input,
output,
gene_column_name='Gene.refGene',
zygozity_column_name='Otherinfo',
delim=','):
""" extract a part of the annotated table that contains candidates for recessive disorders """
table_in = open(input, 'r')
# get right column indexes
header = quote_aware_split(table_in.readline().strip(), delim)
gene_col_index = header.index(gene_column_name)
zygozity_col_index = header.index(zygozity_column_name)
variant_records_per_gene = {}
for l in table_in.xreadlines():
lsplit = quote_aware_split(l, delim)
gene = lsplit[gene_col_index].strip('"')
# the gene record can be a list (e.g. overlapping genes), so it needs to be split
genes = [gene]
if gene.find(',') >= 0:
genes = parenthesis_aware_split(gene, delim=',')
genes = [parenthesis_aware_split(gene, delim=';') for gene in genes]
genes = set([gene for sublist in genes for gene in sublist
]) # unlist and get unique gene ids only
for gene in genes:
# if present, remove suffix in parenthesis
if gene.find('(') >= 0: gene = gene[:gene.find('(')]
# put the variant record in the map
try:
variant_records_per_gene[gene] += [l]
except KeyError:
variant_records_per_gene[gene] = [l]
table_in.close()
# write the table with candidates for recessive inheritance model
table_out = open(output, 'w')
table_out.write(delim.join(header) + '\n')
# iterate over the genes and select...
for gene in variant_records_per_gene:
# ...these with 2 or more variants...
if len(variant_records_per_gene[gene]) >= 2:
for l in variant_records_per_gene[gene]:
table_out.write(l)
# or homozygous variants
else:
lsplit = quote_aware_split(variant_records_per_gene[gene][0])
if lsplit[zygozity_col_index].find('"hom\t') >= 0:
table_out.write(variant_records_per_gene[gene][0])
table_out.close()
def extract_recessive_disorder_candidates(
input, output, gene_column_name="Gene.refGene", zygozity_column_name="Otherinfo", delim=","
):
""" extract a part of the annotated table that contains candidates for recessive disorders """
table_in = open(input, "r")
# get right column indexes
header = quote_aware_split(table_in.readline().strip(), delim)
gene_col_index = header.index(gene_column_name)
zygozity_col_index = header.index(zygozity_column_name)
variant_records_per_gene = {}
for l in table_in.xreadlines():
lsplit = quote_aware_split(l, delim)
gene = lsplit[gene_col_index].strip('"')
# the gene record can be a list (e.g. overlapping genes), so it needs to be split
genes = [gene]
if gene.find(",") >= 0:
genes = parenthesis_aware_split(gene, delim=",")
genes = [parenthesis_aware_split(gene, delim=";") for gene in genes]
genes = set([gene for sublist in genes for gene in sublist]) # unlist and get unique gene ids only
for gene in genes:
# if present, remove suffix in parenthesis
if gene.find("(") >= 0:
gene = gene[: gene.find("(")]
# put the variant record in the map
try:
variant_records_per_gene[gene] += [l]
except KeyError:
variant_records_per_gene[gene] = [l]
table_in.close()
# write the table with candidates for recessive inheritance model
table_out = open(output, "w")
table_out.write(delim.join(header) + "\n")
# iterate over the genes and select...
for gene in variant_records_per_gene:
# ...these with 2 or more variants...
if len(variant_records_per_gene[gene]) >= 2:
for l in variant_records_per_gene[gene]:
table_out.write(l)
# or homozygous variants
else:
lsplit = quote_aware_split(variant_records_per_gene[gene][0])
if lsplit[zygozity_col_index].find('"hom\t') >= 0:
table_out.write(variant_records_per_gene[gene][0])
table_out.close()
def find_geneset_hits_in_samples(sample_files, geneset, gene_column_name = 'Gene.refGene', delim=',', has_header=True, fields=None):
"""
Iterate over the per-sample variant tables (csv, tsv) and look for genes from the geneset.
The gene column is given in the gene_column_name argument: either by name (if has_header==True), or by index of the column (if has_header==False).
Returns a dict in form {sample: {gene1:[variant_line], gene2:[variant_line1, variant_line2]}}, where gene1 and gene2 belong to the geneset,
and variant_line is the entire line from csv file.
As the same gene can appear twice with distinct variants, the dict[sample][gene] is a list.
"""
map={s:{} for s in sample_files}
for fname in sample_files:
sys.stderr.write('Processing '+ fname + '...')
f = open(fname)
gene_col_index = -1
if has_header:
header = quote_aware_split(f.readline().strip(), delim)
gene_col_index = header.index(gene_column_name)
else:
gene_col_index = int(gene_column_name)
for l in f.xreadlines():
gene_entry = quote_aware_split(l, delim)[gene_col_index]
gene_entry = gene_entry.strip().strip('"') # clean the gene name
genes=[gene_entry]
if gene_entry.find(',') >= 0:
genes = parenthesis_aware_split(gene_entry,delim=',') # split multi gene entries
genes = [parenthesis_aware_split(gene,delim=';') for gene in genes]
genes = set([gene for sublist in genes for gene in sublist]) # get unique gene ids only
for gene in genes:
if gene.find('(') > 0:
gene = gene[:gene.find('(')] # strip the transcript change in parenthesis
if gene in geneset:
# clean the records
variant_record = '\t'.join([e.strip().strip('"') for e in quote_aware_split(l, delim)])
# if requested, select a subset of fields
if fields != None:
variant_record = '\t'.join([variant_record.split('\t')[i] for i in fields])
try:
map[fname][gene] += [variant_record]
except KeyError:
map[fname][gene] = [variant_record]
f.close()
sys.stderr.write('done\n')
return map
def include_omim_phenotype_annotation(inputs,
output_table,
gene_column=7,
omim_column=15,
delim=','):
""" include OMIM phenotype into the annotation table """
table_in = open(inputs[1], 'r')
table_out = open(output_table, 'w')
# header
header_in = quote_aware_split(table_in.readline(), delim)
if omim_column <= 0:
omim_column = len(header_in) + 1
header_out = header_in[:omim_column - 1] + ['omim_phenotype'
] + header_in[omim_column - 1:]
table_out.write(delim.join(header_out))
# the rest of the table
for l in table_in.xreadlines():
lsplit = quote_aware_split(l, delim)
gene = lsplit[gene_column - 1].strip('"')
# the gene record can be a list (e.g. overlapping genes), so it needs to be split
genes = [gene]
if gene.find(',') >= 0:
genes = parenthesis_aware_split(gene, delim=',')
genes = [parenthesis_aware_split(gene, delim=';') for gene in genes]
genes = set([gene for sublist in genes for gene in sublist
]) # unlist and get unique gene ids only
for gene in genes:
# if present, remove suffix in parenthesis
if gene.find('(') >= 0:
gene = gene[:gene.find('(')]
# put the variant record in the map
try:
omim_phenotype = omim_gene_phenotype_map[gene]
except KeyError:
omim_phenotype = 'NA'
table_out.write(
delim.join(lsplit[:omim_column - 1] +
['"' + omim_phenotype + '"'] +
lsplit[omim_column - 1:]))
table_in.close()
table_out.close()
def include_omim_phenotype_annotation(inputs, output_table, gene_column=7, omim_column=15, delim=","):
""" include OMIM phenotype into the annotation table """
table_in = open(inputs[1], "r")
table_out = open(output_table, "w")
# header
header_in = quote_aware_split(table_in.readline(), delim)
if omim_column <= 0:
omim_column = len(header_in) + 1
header_out = header_in[: omim_column - 1] + ["omim_phenotype"] + header_in[omim_column - 1 :]
table_out.write(delim.join(header_out))
# the rest of the table
for l in table_in.xreadlines():
lsplit = quote_aware_split(l, delim)
gene = lsplit[gene_column - 1].strip('"')
# the gene record can be a list (e.g. overlapping genes), so it needs to be split
genes = [gene]
if gene.find(",") >= 0:
genes = parenthesis_aware_split(gene, delim=",")
genes = [parenthesis_aware_split(gene, delim=";") for gene in genes]
genes = set([gene for sublist in genes for gene in sublist]) # unlist and get unique gene ids only
for gene in genes:
# if present, remove suffix in parenthesis
if gene.find("(") >= 0:
gene = gene[: gene.find("(")]
# put the variant record in the map
try:
omim_phenotype = omim_gene_phenotype_map[gene]
except KeyError:
omim_phenotype = "NA"
table_out.write(
delim.join(lsplit[: omim_column - 1] + ['"' + omim_phenotype + '"'] + lsplit[omim_column - 1 :])
)
table_in.close()
table_out.close()