def init_associations(self, hdr_only=False):
"""Read GPAD file. HTTP address okay. GZIPPED/BZIPPED file okay."""
associations = []
if self.filename is None:
return associations
ver = None
ntgpadobj = None
hdrobj = GpadHdr()
ifstrm = nopen(self.filename)
for line in ifstrm:
# Read header
if ntgpadobj is None:
if line[0] == '!':
if ver is None and line[1:13] == 'gpa-version:':
ver = line[13:].strip()
hdrobj.chkaddhdr(line)
else:
self.hdr = hdrobj.get_hdr()
if hdr_only:
return associations
ntgpadobj = cx.namedtuple("ntgpadobj", " ".join(self.gpad_columns[ver]))
# Read data
if ntgpadobj is not None:
flds = self._split_line(line)
ntgpad = self._get_ntgpad(ntgpadobj, flds)
associations.append(ntgpad)
# GPAD file has been read
readmsg = " READ {N:7,} associations: {FIN}\n"
sys.stdout.write(readmsg.format(N=len(associations), FIN=self.filename))
return associations
def read_gaf(self, fin_gaf, hdr_only, prt):
"""Read GAF file. HTTP address okay. GZIPPED/BZIPPED file okay."""
ga_lst = []
ver = None
ntgafobj = None
exp_numcol = None
hdrobj = GafHdr()
ifstrm = nopen(fin_gaf)
for line in ifstrm:
# Read header
if ntgafobj is None:
if line[0] == '!':
if line[1:13] == 'gaf-version:':
ver = line[13:].strip()
hdrobj.chkaddhdr(line)
else:
self.hdr = hdrobj.get_hdr()
if hdr_only:
return ga_lst
ntgafobj = cx.namedtuple("ntgafobj",
" ".join(self.gaf_columns[ver]))
exp_numcol = self.gaf_numcol[ver]
# Read data
if ntgafobj is not None:
flds = self._split_line(line, exp_numcol)
ntgaf = self._get_ntgaf(ntgafobj, flds, ver)
ga_lst.append(ntgaf)
# GAF file has been read
if prt is not None:
readmsg = " READ {N:,} associations: {FIN}\n"
prt.write(readmsg.format(N=len(ga_lst), FIN=fin_gaf))
return self.evobj.sort_nts(ga_lst, 'Evidence_Code')
def init_associations(self, hdr_only=False, namespaces=None, prt=sys.stdout):
"""Read GPAD file. HTTP address okay. GZIPPED/BZIPPED file okay."""
import timeit
import datetime
associations = []
tic = timeit.default_timer()
if self.filename is None:
return associations
ver = None
ntgpadobj_make = None
hdrobj = GpadHdr()
ifstrm = nopen(self.filename)
_add_ns = self.godag is not None
_get_ntgpadvals = self._get_ntgpadvals
get_all_nss = self._get_b_all_nss(namespaces)
for lnum, line in enumerate(ifstrm, 1):
# Read data
if ntgpadobj_make:
flds = self._split_line(line)
try:
# pylint: disable=not-callable
goid = flds[3]
nspc = self._get_namespace(goid) if _add_ns else None
if get_all_nss or nspc in namespaces:
ntgpad = ntgpadobj_make(_get_ntgpadvals(flds, goid, nspc, _add_ns))
associations.append(ntgpad)
# pylint: disable=broad-except
except Exception as inst:
import traceback
traceback.print_exc()
sys.stdout.write("\n **FATAL: {MSG}\n\n".format(MSG=str(inst)))
sys.stdout.write("**FATAL: {FIN}[{LNUM}]:\n{L}\n".format(
FIN=self.filename, L=line, LNUM=lnum))
for idx, (key, val) in enumerate(zip(self.gpadhdr, flds)):
sys.stdout.write('{I:2} {KEY:13} {VAL}\n'.format(I=idx, KEY=key, VAL=val))
## if datobj is not None:
## datobj.prt_line_detail(sys.stdout, line)
sys.exit(1)
# Read header
else:
if line[0] == '!':
if ver is None and line[1:13] == 'gpa-version:':
ver = line[13:].strip()
hdrobj.chkaddhdr(line)
else:
self.hdr = hdrobj.get_hdr()
if hdr_only:
return associations
ntgpadobj_make = self._get_ntgpadnt(ver, _add_ns)._make
# GPAD file has been read
prt.write('HMS:{HMS} {N:7,} annotations READ: {ANNO} {NSs}\n'.format(
N=len(associations), ANNO=self.filename,
NSs=','.join(namespaces) if namespaces else '',
HMS=str(datetime.timedelta(seconds=(timeit.default_timer()-tic)))))
return associations
def init_associations(self, hdr_only=False, namespaces=None, prt=sys.stdout):
"""Read GPAD file. HTTP address okay. GZIPPED/BZIPPED file okay."""
import timeit
import datetime
associations = []
tic = timeit.default_timer()
if self.filename is None:
return associations
ver = None
ntgpadobj_make = None
hdrobj = GpadHdr()
ifstrm = nopen(self.filename)
_add_ns = self.godag is not None
_get_ntgpadvals = self._get_ntgpadvals
get_all_nss = self._get_b_all_nss(namespaces)
for lnum, line in enumerate(ifstrm, 1):
# Read data
if ntgpadobj_make:
flds = self._split_line(line)
try:
# pylint: disable=not-callable
goid = flds[3]
nspc = self._get_namespace(goid) if _add_ns else None
if get_all_nss or nspc in namespaces:
ntgpad = ntgpadobj_make(_get_ntgpadvals(flds, goid, nspc, _add_ns))
associations.append(ntgpad)
# pylint: disable=broad-except
except Exception as inst:
import traceback
traceback.print_exc()
sys.stdout.write("\n **FATAL: {MSG}\n\n".format(MSG=str(inst)))
sys.stdout.write("**FATAL: {FIN}[{LNUM}]:\n{L}\n".format(
FIN=self.filename, L=line, LNUM=lnum))
for idx, val in enumerate(flds):
sys.stdout.write('{I:2} {VAL}\n'.format(I=idx, VAL=val))
## if datobj is not None:
## datobj.prt_line_detail(sys.stdout, line)
sys.exit(1)
# Read header
else:
if line[0] == '!':
if ver is None and line[1:13] == 'gpa-version:':
ver = line[13:].strip()
hdrobj.chkaddhdr(line)
else:
self.hdr = hdrobj.get_hdr()
if hdr_only:
return associations
ntgpadobj_make = self._get_ntgpadnt(ver, _add_ns)._make
# GPAD file has been read
prt.write('HMS:{HMS} {N:7,} annotations READ: {ANNO} {NSs}\n'.format(
N=len(associations), ANNO=self.filename,
NSs=','.join(namespaces) if namespaces else '',
HMS=str(datetime.timedelta(seconds=(timeit.default_timer()-tic)))))
return associations
def read_gaf(self, fin_gaf):
"""Read GAF file. HTTP address okay. GZIPPED/BZIPPED file okay."""
ga_lst = []
ifstrm = nopen(fin_gaf)
ver = None
ntgafobj = None
exp_numcol = None
for line in ifstrm:
if ntgafobj is not None and not line.startswith('!'):
flds = self._split_line(line, exp_numcol)
ntgaf = self._get_ntgaf(ntgafobj, flds, ver)
ga_lst.append(ntgaf)
elif ntgafobj is None and line.startswith('!gaf-version:'):
ver = line[13:].strip()
ntgafobj = namedtuple("ntgafobj", " ".join(self.gaf_columns[ver]))
exp_numcol = self.gaf_numcol[ver]
self.log.write(" READ {N:,} items: {FIN}\n".format(N=len(ga_lst), FIN=fin_gaf))
return ga_lst
def read_gaf(self, fin_gaf):
"""Read GAF file. HTTP address okay. GZIPPED/BZIPPED file okay."""
ga_lst = []
ifstrm = nopen(fin_gaf)
ver = None
ntgafobj = None
exp_numcol = None
for line in ifstrm:
if ntgafobj is not None and not line.startswith("!"):
flds = self._split_line(line, exp_numcol)
ntgaf = self._get_ntgaf(ntgafobj, flds, ver)
ga_lst.append(ntgaf)
elif ntgafobj is None and line.startswith("!gaf-version:"):
ver = line[13:].strip()
ntgafobj = cx.namedtuple("ntgafobj", " ".join(self.gaf_columns[ver]))
exp_numcol = self.gaf_numcol[ver]
self.log.write(" READ {N:,} associations: {FIN}\n".format(N=len(ga_lst), FIN=fin_gaf))
ga_lst = self.evobj.sort_nts(ga_lst, "Evidence_Code")
return ga_lst
