def test_get_default_config(self):
# Initialization
cc = ConfigClient()
cbc = CellBaseClient(cc)
assert cc.get_default_configuration() == cbc.get_default_configuration(
)
def test_get(self):
""""Checks generic fetcher for RESTful service"""
# Initialization
cc = ConfigClient()
cbc = CellBaseClient(cc)
res = cbc.get('feature', 'gene', 'protein', 'BRCA1')
assert res[0]['result'][0]['name'][0] == 'BRCA1_HUMAN'
def test_get(self):
""""Checks generic fetcher for RESTful service"""
# Initialization
cc = ConfigClient()
cbc = CellBaseClient(cc)
res = cbc.get('feature', 'gene', 'protein', 'BRCA1')
assert res[0]['result'][0]['name'][0] == 'BRCA1_HUMAN'
def test_change_config(self):
"""Checks configuration customization"""
# Initialization
cc = ConfigClient()
cbc = CellBaseClient(cc)
# Checking some default config params
assert cbc.get_config()['species'] == 'hsapiens'
assert cbc.get_config()['version'] == 'v4'
# Checking some setters for config params
cc.species = 'mmusculus'
assert cbc.get_config()['species'] == 'mmusculus'
cc.version = 'v3'
assert cbc.get_config()['version'] == 'v3'
def test_change_config(self):
"""Checks configuration customization"""
# Initialization
cc = ConfigClient()
cbc = CellBaseClient(cc)
# Checking some default config params
assert cbc.get_config()['species'] == 'hsapiens'
assert cbc.get_config()['version'] == 'latest'
# Checking some setters for config params
cc.species = 'mmusculus'
assert cbc.get_config()['species'] == 'mmusculus'
cc.version = 'v4'
assert cbc.get_config()['version'] == 'v4'
def main():
"""The main function"""
# Getting args
args = _parse_arguments()
# Check arguments
args = _check_arguments(args)
# Setting up logging system
_set_logger(args.verbosity)
# Setting up PyCellBase clients
cc = ConfigClient({
"species": _DEFAULT_SPECIES,
"version": _DEFAULT_API_VERSION,
"rest": {
"hosts": [_DEFAULT_HOST]
}
})
# Overriding config
if args.config is not None:
cc = ConfigClient(args.config)
if args.species is not None:
cc.species = args.species
if args.api_version is not None:
cc.version = args.api_version
if args.host is not None:
cc.host = args.host
if args.assembly is not None:
assembly = args.assembly
else:
assembly = _DEFAULT_ASSEMBLY
cbc = CellBaseClient(cc)
if args.which == 'xref':
# Getting available databases
databases = _get_databases_list(cbc, assembly)
# Filtering databases with include and exclude
include = args.include.split(',') if args.include is not None else None
exclude = args.exclude.split(',') if args.exclude is not None else None
databases = _filter_databases(databases,
include=include,
exclude=exclude)
# Converting IDs
convert_ids(input_fpath=args.input_fpath,
output_fpath=args.output_fpath,
cellbase_client=cbc,
assembly=assembly,
databases=databases)
if args.which == 'hgvs':
calculate_hgvs(input_data=args.input,
output_fpath=args.output_fpath,
cbc=cbc,
ref_seq_type=args.ref_seq_type,
assembly=assembly)
if args.which == 'annotate':
# Getting the list of annotations to add
include = get_include_annots(args.include, args.exclude)
include = list(set(include).intersection(_ANNOT.keys()))
# Annotate VCF
annotate_vcf(cellbase_client=cbc,
input_fpath=args.input_fpath,
output_fpath=args.output_fpath,
include=include,
assembly=assembly)
#!/usr/bin/env python2
# coding: utf-8
import re
from os import listdir
from os.path import isfile, join
from pycellbase.cbclient import CellBaseClient
cbc = CellBaseClient()
cbcregion = cbc.get_genomic_region_client()
def GetConservation(region):
query = str(region[0]) + ":" + str(region[1]) + "-" + str(region[2])
conservation = cbcregion.get_conservation(query, assembly="GRCh38")
for result in conservation[0]['result']:
if 'phastCons' in result['source']:
phastCons = result['values']
if 'phylop' in result['source']:
phylop = result['values']
positions = [str(i) for i in range(region[1], region[2], 1)]
phastCons_coverage = float(len(phastCons)) / float(len(positions))
phylop_coverage = float(len(phylop)) / float(len(positions))
return phastCons, phylop, phastCons_coverage, phylop_coverage
def GetPhastCons_old(region, PhastConscores):
positions = [
str(region[0]) + '_' + str(i) for i in range(region[1], region[2], 1)
]
scores = [
float(PhastConscores[position]) for position in positions