def _setup(self):
self.infile = pysam.TabixFile(self.filename)
contig_name = self.infile.contigs[-1]
self._has_chrom_prefix = contig_name.startswith("chr")
self._region_reset(self.region)
self.schema = Schema.from_dict({"str": self._header_read()})
def _setup(self):
assert os.path.exists(self.filename)
assert self.is_gzip(self.filename)
self.infile = gzip.open(self.filename, "rt")
if self.options.vcf:
self._skip_metalines()
self.schema = Schema.from_dict({"str": self._header_read()})
def _setup(self):
if self.filename == "-":
self.infile = sys.stdin
else:
assert os.path.exists(self.filename), self.filename
assert not self.is_gzip(self.filename)
self.infile = open(self.filename, "r")
if self.options.vcf:
self._skip_metalines()
self.schema = Schema.from_dict({"str": self._header_read()})
def test_schema_from_config(sample_schema_dict):
expected_columns = {
"chr": "str",
"position": "str",
"variant": "str",
"dummy_score": "float",
}
schema = Schema.from_dict(sample_schema_dict)
for col, type_ in expected_columns.items():
assert col in schema.columns
assert type_ == schema.columns[col].type_name
def test_merge_schemas(generic_schema, generic_schema_alt):
schema = Schema.merge_schemas(generic_schema, generic_schema_alt)
expected_cols = [
"col1",
"col2",
"col3",
"col4",
"col5",
"col6",
"col7",
"col8",
"col11",
"col12",
]
assert list(schema.columns.keys()) == expected_cols
def __init__(self, score_filename, config_filename=None):
self.score_filename = score_filename
assert os.path.exists(self.score_filename), self.score_filename
if config_filename is None:
config_filename = "{}.conf".format(self.score_filename)
self.config = GPFConfigParser.load_config(config_filename,
score_file_conf_schema)
assert self.config.general.header is not None
assert self.config.columns.score is not None
self.header = self.config.general.header
logger.debug(f"score file {os.path.basename(self.score_filename)} "
f"header {self.header}")
self.score_names = self.config.columns.score
self.schema = Schema.from_dict(self.config.score_schema).order_as(
self.header)
logger.debug(f"score file {os.path.basename(self.score_filename)} "
f"schema {self.schema.col_names}")
assert all([sn in self.schema for sn in self.score_names]), [
self.score_filename,
self.score_names,
self.schema.col_names,
]
self.chr_index = self.schema.col_names.index(self.chr_name)
self.pos_begin_index = self.schema.col_names.index(self.pos_begin_name)
self.pos_end_index = self.schema.col_names.index(self.pos_end_name)
self.chr_prefix = getattr(self.config.general, "chr_prefix", False)
self.no_score_value = self.config.general.no_score_value or "na"
if self.no_score_value.lower() in ("na", "none"):
self.no_score_value = None
self._init_access()
def merge_schemas(cls, left, right):
return cls.convert(Schema.merge_schemas(left, right))
def from_dict(cls, schema_dict):
return cls.convert(Schema.from_dict(schema_dict))
def generic_schema_alt():
return Schema.from_dict({
"str": ["col1", "col7", "col8"],
"float": ["col11", "col12", "col6"]
})
def generic_schema():
return Schema.from_dict({
"str": ["col1", "col2", "col3"],
"float": ["col4", "col5", "col6"]
})