def check_hgvs(parser, validator, hdp, hgvscode, jsonfile):
"""Checks the HGVS code. Prints error and the HGVS code if HGVS is not recognized and saves
the error information in the overview directory
Args:
parser (hgvs.parser.Parser object): Used to parse the HGVS code
validator (hgvs.validator.Validator object): Used to validate the parsed HGVS code
hgvs (string): Input that is going to be tested
Returns:
correct (boolean): True if HGVS-Code is correct
"""
try:
variant = parser.parse_hgvs_variant(hgvscode)
except hgvs.exceptions.HGVSError as e:
if "char 1: expected a letter or digit" in e:
return (errorcorrbymut(hgvscode, jsonfile))
vm = hgvs.assemblymapper.AssemblyMapper(hdp,
assembly_name='GRCh37',
alt_aln_method='splign')
var_g = vm.c_to_g(variant)
try:
correct = validator.validate(hp.parse_hgvs_variant(str(variant)))
except Exception:
return (False)
return (True)
def check_hgvs(parser, validator, hgvs, submitterteam, submitter, overview, file):
"""Checks the HGVS code. Prints error and the HGVS code if HGVS is not recognized and saves
the error information in the overview directory
Args:
parser (hgvs.parser.Parser object): Used to parse the HGVS code
validator (hgvs.validator.Validator object): Used to validate the parsed HGVS code
hgvs (string): Input that is going to be tested
submitterteam (string): Team of submitters in the overview dictionary
submitter (string): Name of submitter
overview (dictionary): Dictionary containing the information of previously parsed JSONs
file (string): directory of json-file
"""
try:
hgvstest = validator.validate(parser.parse_hgvs_variant(hgvs))
except Exception:
if file not in overview[submitterteam]['incorrect JSONs']:
overview[submitterteam]['incorrect JSONs'][file] = {}
if 'falscher HGVS-Code' not in overview[submitterteam]['incorrect JSONs'][file]:
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'] = {}
overview[submitterteam]['incorrect JSONs'][file]['submitter'] = submitter
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'][hgvs] = "Parsing error"
else:
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'][hgvs] = "Parsing error"
if hgvstest == False:
if file not in overview[submitterteam]['incorrect JSONs']:
overview[submitterteam]['incorrect JSONs'][file] = {}
if 'falscher HGVS-Code' not in overview[submitterteam]['incorrect JSONs'][file]:
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'] = {}
overview[submitterteam]['incorrect JSONs'][file]['submitter'] = submitter
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'][hgvs] = "internally incorrect HGVS"
else:
overview[submitterteam]['incorrect JSONs'][file]['falscher HGVS-Code'][hgvs] = "internally incorrect HGVS"
def parsehgvs(hgvs, parser):
parsedhgvs = parser.parse_hgvs_variant(hgvs)
chrom = str(int(parsedhgvs.ac.split(".")[0][-2:]))
offset = parsedhgvs.posedit.pos.start.offset
ref = parsedhgvs.posedit.ref
alt = parsedhgvs.posedit.alt
return (chrom, offset, ref, alt)
def from_hgvs(SQ, hgvs_expr):
parser = hgvs.parser.Parser()
sv = parser.parse_hgvs_variant(hgvs_expr)
if isinstance(sv.posedit.pos, hgvs.location.BaseOffsetInterval):
if sv.posedit.pos.start.is_intronic or sv.posedit.pos.end.is_intronic:
raise ValueError(
"Intronic HGVS variants are not supported ({sv.posedit})")
if sv.posedit.edit.type == 'ins':
start = sv.posedit.pos.start.base
end = sv.posedit.pos.start.base
state = sv.posedit.edit.alt
elif sv.posedit.edit.type in ('sub', 'del', 'delins', 'identity'):
start = sv.posedit.pos.start.base - 1
end = sv.posedit.pos.end.base
state = sv.posedit.edit.alt or ''
else:
raise ValueError(
f"HGVS variant type {sv.posedit.edit.type} is unsupported")
location = identifiers.getVSL(SQ, start, end)
vsl = 'ga4gh:VSL.' + identifiers.digest(
bytes(json.dumps(location), 'utf-8'))
allele = identifiers.getVA(vsl, state)
va = 'ga4gh:VA.' + identifiers.digest(bytes(json.dumps(allele), 'utf-8'))
model = identifiers.assembleJSON([SQ, sv.ac], location, allele, va)
return json.dumps(model, ensure_ascii=False, indent=4)
def parse_hgvs(parser, candidate):
assert candidate
# Normalize single deletions: NM_007300.3:c.1100C>None -> NM_007300.3:c.1100delC
candidate = re.sub(r"(NM.*c\.\d*)([ATCGatcg]+)(>None)", r"\1del\2", candidate)
# TODO: Normalize multiple deletions and delins
# TODO: Limit to only specific BRCA transcripts
# ex: 23199084 NM_007294.3:c.2681AA>None|NM_007300.3:c.2681AA>None
try:
return str(parser.parse_hgvs_variant(candidate))
except hgvs.exceptions.HGVSParseError:
print("Failed to parse: {}".format(candidate))
return ""
def normalize_variants(variants, parser, mapper):
# Generate normalized genomic and coding hgvs strings
# REMIND: This NC_0000<chr>.11 is a hack, will not work correctly
# for single digit chromosomes...
variants["norm_g_hgvs"] = variants.apply(
lambda row: "NC_0000{}.11:g.{}{}>{}".format(
row.Chr, row.Pos, row.Ref, row.Alt) if (len(row.Ref) == 1) and
(len(row.Alt) == 1) else "NC_0000{}.11:g.{}_{}del{}ins{}".format(
row.Chr, row.Pos, row.Pos + len(row.Ref), row.Ref, row.Alt),
axis="columns")
# Normalize the hgvs string by parsing via the hgvs package
variants["norm_g_hgvs"] = variants.apply(
lambda row: str(parser.parse_hgvs_variant(row.norm_g_hgvs)),
axis="columns")
return variants
def hgvs_c_to_g(candidate, parser, mapper):
"""
Try to parse candidate hgvs coding string from a paper and if
successful try and map it to an hgvs genomic string.
Returns the parsed coding and mapped genomic hgvs
"""
print(candidate)
# Normalize single deletions: NM_007300.3:c.1100C>None -> NM_007300.3:c.1100del
# candidate = re.sub(r"(NM.*c\.\d*)([ATCG]>None)", r"\1del", candidate)
# Normalize single deletions: NM_007300.3:c.1100C>None -> NM_007300.3:c.1100delC
candidate = re.sub(r"(NM.*c\.\d*)([ATCG])>None", r"\1del\2", candidate)
# TODO: Normalize multiple deletions and delins
# TODO: Limit to only specific BRCA transcripts
# ex: 23199084 NM_007294.3:c.2681AA>None|NM_007300.3:c.2681AA>None
print("=> {}".format(candidate))
# Try to parse and map back to genomic
try:
parsed_hgvs = parser.parse_hgvs_variant(candidate)
print("=> {}".format(parsed_hgvs))
try:
if parsed_hgvs.type == "c":
norm_g_hgvs = mapper.c_to_g(parsed_hgvs)
print("=> {}".format(norm_g_hgvs))
return str(parsed_hgvs), str(norm_g_hgvs)
else:
print("Only coding variants (.c) supported")
except hgvs.exceptions.HGVSInvalidVariantError:
print("Failed Mapping: HGVSInvalidVariantError")
except hgvs.exceptions.HGVSInvalidIntervalError:
print("Failed Mapping: HGVSInvalidIntervalError")
except hgvs.exceptions.HGVSParseError:
print("Failed Parsing")
return "", ""