def concatenate_vcf(infiles, outfile):
'''
Concatenate VCF files
:param infiles: dictionary of input VCF files to be concatenated
:param outfile: output VCF file
'''
if isinstance(infiles, dict):
keys = infiles.keys()
keys = sorted(keys)
infiles = [infiles[val] for val in keys]
with helpers.GetFileHandle(outfile, 'w') as ofile:
header = None
for ifile in infiles:
if os.path.getsize(ifile) == 0:
warnings.warn('input file {} is empty'.format(ifile))
continue
with helpers.GetFileHandle(ifile) as f:
if not header:
header = _get_header(f)
for line in header:
ofile.write(line)
else:
if not _get_header(f) == header:
warnings.warn(
'merging vcf files with mismatching headers')
for l in f:
print >> ofile, l,
def write_csv_with_header(self, infile, headerless_input=True):
with helpers.GetFileHandle(self.filepath, 'wt') as writer:
with helpers.GetFileHandle(infile) as reader:
if headerless_input:
writer.write(self.header_line)
self.write_csv_data(reader, writer)
self.__write_yaml()
def write_headerless_csv(self, infile):
with helpers.GetFileHandle(self.filepath, 'wt') as writer:
with helpers.GetFileHandle(infile) as reader:
if not reader.readline() == self.header_line:
raise CsvWriterError("cannot write, wrong header")
self.write_csv_data(reader, writer)
self.__write_yaml()
def concatenate_files(self, infiles):
header = self.header_line if self.header else None
with helpers.GetFileHandle(self.filepath, 'wt') as writer:
if header:
writer.write(header)
for infile in infiles:
with helpers.GetFileHandle(infile) as reader:
self.write_csv_data(reader, writer)
self.__write_yaml()
def generate_metadata(self):
with helpers.GetFileHandle(self.filepath) as inputfile:
header = inputfile.readline().strip()
sep = self.__detect_sep_from_header(header)
columns = header.split(sep)
header = True
dtypes = self.__generate_dtypes(sep=sep)
return header, sep, dtypes, columns
def __write_yaml(self):
yamldata = {'header': self.header, 'sep': self.sep, 'columns': []}
for column in self.columns:
data = {'name': column, 'dtype': self.dtypes[column]}
yamldata['columns'].append(data)
with helpers.GetFileHandle(self.yaml_file, 'wt') as f:
yaml.safe_dump(yamldata, f, default_flow_style=False)
def __parse_metadata(self):
with helpers.GetFileHandle(self.filepath + '.yaml') as yamlfile:
yamldata = yaml.safe_load(yamlfile)
header = yamldata['header']
sep = yamldata.get('sep', ',')
dtypes = {}
columns = []
for coldata in yamldata['columns']:
colname = coldata['name']
dtypes[colname] = coldata['dtype']
columns.append(colname)
return header, sep, dtypes, columns
def __detect_sep_from_file(self):
with helpers.GetFileHandle(self.filepath) as reader:
header = reader.readline().strip()
return self.__detect_sep_from_header(header)
