def dnld_assc(assc_name, go2obj, prt=sys.stdout):
"""Download association from http://geneontology.org/gene-associations."""
# Example assc_name: "tair.gaf"
# Download the Association
dirloc, assc_base = os.path.split(assc_name)
if not dirloc:
dirloc = os.getcwd()
assc_locfile = os.path.join(dirloc, assc_base) if not dirloc else assc_name
if not os.path.isfile(assc_locfile):
# assc_http = "http://geneontology.org/gene-associations/"
assc_http = "http://current.geneontology.org/annotations/"
for ext in ['gz']:
src = os.path.join(assc_http, "{ASSC}.{EXT}".format(ASSC=assc_base,
EXT=ext))
dnld_file(src, assc_locfile, prt, loading_bar=None)
# Read the downloaded association
assc_orig = read_gaf(assc_locfile, prt)
if go2obj is None:
return assc_orig
# If a GO DAG is provided, use only GO IDs present in the GO DAG
assc = {}
goids_dag = set(go2obj.keys())
for gene, goids_cur in assc_orig.items():
assc[gene] = goids_cur.intersection(goids_dag)
return assc
def dnld_goa(self, species, ext='gaf', item=None, fileout=None):
"""Download GOA source file name on EMBL-EBI ftp server."""
basename = self.get_basename(species, ext, item)
src = os.path.join(self.ftp_src_goa, species.upper(), "{F}.gz".format(F=basename))
dst = os.path.join(os.getcwd(), basename) if fileout is None else fileout
dnld_file(src, dst, prt=sys.stdout, loading_bar=None)
return dst
def dnld_annotation(assc_file, prt=sys.stdout):
"""Download gaf, gpad, or gpi from http://current.geneontology.org/annotations/"""
if not os.path.isfile(assc_file):
# assc_http = "http://geneontology.org/gene-associations/"
assc_http = "http://current.geneontology.org/annotations/"
_, assc_base = os.path.split(assc_file)
src = os.path.join(assc_http, "{ASSC}.gz".format(ASSC=assc_base))
dnld_file(src, assc_file, prt, loading_bar=None)
def dnld_annotation(assc_file, prt=sys.stdout):
"""Download gaf, gpad, or gpi from http://current.geneontology.org/annotations/"""
if not os.path.isfile(assc_file):
# assc_http = "http://geneontology.org/gene-associations/"
assc_http = "http://current.geneontology.org/annotations/"
_, assc_base = os.path.split(assc_file)
src = os.path.join(assc_http, "{ASSC}.gz".format(ASSC=assc_base))
dnld_file(src, assc_file, prt, loading_bar=None)
def dnld_assc(assc_name, go2obj, prt=sys.stdout):
"""Download association from http://geneontology.org/gene-associations."""
# Example assc_name: "gene_association.tair"
dirloc, assc_base = os.path.split(assc_name)
if not dirloc:
dirloc = os.getcwd()
assc_local = os.path.join(dirloc, assc_base) if not dirloc else assc_name
if not os.path.isfile(assc_local):
assc_http = "http://geneontology.org/gene-associations/"
for ext in ['gz']:
src = os.path.join(assc_http, "{ASSC}.{EXT}".format(ASSC=assc_base, EXT=ext))
dnld_file(src, assc_local, prt, loading_bar=None)
assc = {}
goids_dag = set(go2obj.keys())
for gene, goids_cur in read_gaf(assc_local, prt).items():
assc[gene] = goids_cur.intersection(goids_dag)
return assc