def collectCore(self, vcfname, headerKey=None):
"""
Return a data frame with features collected from the given VCF
If headerKey is provided, then use this header value to extract labels
for the INFO EVSF feature tag
"""
feature_labels = ["CHROM", "POS", "REF", "ALT"]
header_feature_labels = None
records = []
isHeader = True
isHeaderKey = (headerKey is not None)
for line in openMaybeGzip(vcfname):
if isHeader:
if line[0] == "#":
if isHeaderKey and line.startswith("##"):
word = line[2:].strip().split("=")
if word[0] == headerKey:
assert (header_feature_labels is None)
header_feature_labels = word[1].split(",")
# print header_feature_labels
assert (len(header_feature_labels) > 0)
continue
else:
if isHeaderKey:
assert (header_feature_labels is not None)
isHeader = False
word = line.strip().split('\t')
qrec = {
"CHROM": word[VCFID.CHROM],
"POS": int(word[VCFID.POS]),
"REF": word[VCFID.REF],
"ALT": word[VCFID.ALT]
}
if isHeaderKey:
for ikv in word[VCFID.INFO].split(';'):
iword = ikv.split("=", 1)
if iword[0] != "EVSF": continue
assert (len(iword) == 2)
if len(word[VCFID.ALT]) > 1: continue # skip indels
features = [float(f) for f in iword[1].split(',')]
# print features
# assert(len(features) == len(header_feature_labels))
for i in range(len(features)):
qrec[header_feature_labels[i]] = features[i]
records.append(qrec)
cols = feature_labels
if isHeaderKey:
cols += header_feature_labels
return pandas.DataFrame(records, columns=cols)
def collectCore(self, vcfname, headerKey = None):
"""
Return a data frame with features collected from the given VCF
If headerKey is provided, then use this header value to extract labels
for the INFO EVSF feature tag
"""
def isNucleotide(nucString):
"""
Return True if nucString is a single nucleotide, False otherwise.
"""
return (nucString in ["A", "C", "G", "T", "N"])
def variantType(ref, alt):
"""
Return 'snv' if ref and all alt alleles are single nucleotides,
otherwise return 'indel'
"""
if isNucleotide(ref) and all(isNucleotide(allele) for allele in alt.split(',')) :
return "snv"
return "indel"
def processVariant(line, keyType, header_feature_labels):
"""
Return a record with collected features for this variant
or None if the variant type does not match the key type.
"""
isHeaderKey = (header_feature_labels is not None)
word = line.strip().split('\t')
qrec = {
"CHROM": word[VCFID.CHROM],
"POS": int(word[VCFID.POS]),
"REF": word[VCFID.REF],
"ALT": word[VCFID.ALT],
}
if isHeaderKey :
if variantType(word[VCFID.REF], word[VCFID.ALT]) != keyType :
return None
for ikv in word[VCFID.INFO].split(';') :
iword = ikv.split("=",1)
if iword[0] == "EVSF" :
assert(len(iword) == 2)
features = [float(f) for f in iword[1].split(',')]
for i in range(len(features)) :
qrec[header_feature_labels[i]] = features[i]
return qrec
feature_labels = ["CHROM", "POS", "REF", "ALT"]
header_feature_labels = None
records = []
isHeader = True
isHeaderKey = (headerKey is not None)
if isHeaderKey :
if headerKey == "snv_scoring_features" :
keyType = "snv"
elif headerKey == "indel_scoring_features" :
keyType = "indel"
else :
raise Exception("Unknown header key: '%s'" % headerKey)
else :
keyType = None
for line in openMaybeGzip(vcfname):
if isHeader :
if line[0] == "#" :
if isHeaderKey and line.startswith("##") :
word = line[2:].strip().split("=")
if word[0] == headerKey :
assert(header_feature_labels is None)
header_feature_labels = word[1].split(",")
assert(len(header_feature_labels) > 0)
continue
else :
if isHeaderKey :
assert(header_feature_labels is not None)
isHeader = False
qrec = processVariant(line, keyType, header_feature_labels)
if qrec is not None:
records.append(qrec)
cols = feature_labels
if isHeaderKey :
cols += header_feature_labels
return pandas.DataFrame(records, columns=cols)