def do_uploadfsa(args, dbh):
cout('Uploading FSA files from input file: %s' % args.infile)
b = dbh.get_batch(args.batch)
inrows = csv.DictReader(
open(args.infile),
delimiter=',' if args.infile.endswith('.csv') else '\t')
#next(inrows)
total_fsa = 0
line_counter = 1
for r in inrows:
line_counter += 1
#if not row[0] or row[0].startswith('#'):
# continue
if not (r['FILENAME'] and r['SAMPLE']) or '#' in [
r['FILENAME'][0], r['SAMPLE'][0]
]:
continue
#if len(row) < 3:
# cerr('ERR - line %d only has %d item(s)' % (line_counter, len(row)))
sample_code, fsa_filename, fsa_panel = r['SAMPLE'], r['FILENAME'], r[
'PANEL']
if r['OPTIONS']:
options = tokenize(r['OPTIONS'])
else:
options = None
try:
s = b.search_sample(sample_code)
if not s:
cerr('ERR - sample %s does not exist' % sample_code)
sys.exit(1)
with open(args.indir + '/' + fsa_filename, 'rb') as f:
trace = f.read()
a = s.add_fsa_assay(trace,
filename=fsa_filename,
panel_code=fsa_panel,
options=options,
species=args.species,
dbhandler=dbh)
cerr('INFO - sample: %s assay: %s panel: %s has been uploaded' %
(s.code, a.filename, fsa_panel))
except Exception as exc:
if not args.test:
raise
cerr('ERR - line %d' % line_counter)
cerr(' => %s' % str(exc))
def do_uploadfsa(args, dbh):
cout('Uploading FSA files from input file: %s' % args.infile)
b = dbh.get_batch(args.batch)
inrows = csv.DictReader( open(args.infile),
delimiter = ',' if args.infile.endswith('.csv') else '\t' )
#next(inrows)
total_fsa = 0
line_counter = 1
for r in inrows:
line_counter += 1
#if not row[0] or row[0].startswith('#'):
# continue
if not (r['FILENAME'] and r['SAMPLE']) or '#' in [ r['FILENAME'][0], r['SAMPLE'][0] ]:
continue
#if len(row) < 3:
# cerr('ERR - line %d only has %d item(s)' % (line_counter, len(row)))
sample_code, fsa_filename, fsa_panel = r['SAMPLE'], r['FILENAME'], r['PANEL']
if r['OPTIONS']:
options = tokenize( r['OPTIONS'] )
else:
options = None
try:
s = b.search_sample(sample_code)
if not s:
cerr('ERR - sample %s does not exist' % sample_code)
sys.exit(1)
with open( args.indir + '/' + fsa_filename, 'rb') as f:
trace = f.read()
a = s.add_fsa_assay( trace, filename=fsa_filename, panel_code = fsa_panel,
options = options, species = args.species, dbhandler = dbh)
cerr('INFO - sample: %s assay: %s panel: %s has been uploaded' %
(s.code, a.filename, fsa_panel))
except Exception as exc:
if not args.test:
raise
cerr('ERR - line %d' % line_counter)
cerr(' => %s' % str(exc))
def open_fsa( args ):
""" open FSA file(s) and prepare fsa instances
requires: args.file, args.panel, args.panelfile
"""
from fatools.lib.fileio.models import Marker, Panel, FSA
if not args.panel:
cexit('ERR: using FSA file(s) requires --panel argument!')
if not args.panelfile:
cerr('WARN: using default built-in panels')
Panel.upload(params.default_panels)
else:
with open(args.panelfile) as f:
# open a YAML file that describe panel sets
Panel.upload(yaml.load(f))
if not args.markerfile:
Marker.upload(params.default_markers)
else:
raise NotImplementedError()
panel = Panel.get_panel(args.panel)
fsa_list = []
index = 1
# prepare caching
cache_path = None
if not args.no_cache:
cache_path = os.path.join(os.path.expanduser('~'), '.fatools_caches', 'channels')
if args.cache_path is not None:
cache_path = os.path.join(args.cache_path, '.fatools_caches', 'channels')
if not os.path.exists(cache_path):
os.makedirs(cache_path)
if args.file:
for fsa_filename in args.file.split(','):
fsa_filename = fsa_filename.strip()
fsa = FSA.from_file(fsa_filename, panel, cache=not args.no_cache,
cache_path=cache_path)
# yield (fsa, str(i))
fsa_list.append( (fsa, str(index)) )
index += 1
elif args.infile:
with open(args.infile) as f:
buf, delim = detect_buffer( f.read() )
inrows = csv.DictReader( StringIO(buf), delimiter=delim )
line = 1
index = 1
for r in inrows:
line += 1
fsa_filename = r['FILENAME'].strip()
if fsa_filename.startswith('#'):
continue
if r.get('OPTIONS', None):
options = tokenize( r['OPTIONS'] )
else:
options = None
panel_code = r.get('PANEL', None) or args.panel
panel = Panel.get_panel(panel_code)
fsa = FSA.from_file(fsa_filename, panel, options, cache=not args.no_cache,
cache_path=cache_path)
if 'SAMPLE' in inrows.fieldnames:
# yield (fsa, r['SAMPLE'])
fsa_list.append( (fsa, r['SAMPLE']) )
else:
# yield (fsa, str(index))
fsa_list.append( (fsa, str(index)) )
index += 1
return fsa_list
def open_fsa(args, _params):
""" open FSA file(s) and prepare fsa instances
requires: args.file, args.panel, args.panelfile
"""
from fatools.lib.fileio.models import Marker, Panel, FSA
if not args.panel:
cexit('ERR: using FSA file(s) requires --panel argument!')
if not args.panelfile:
cerr('WARN: using default built-in panels')
Panel.upload(params.default_panels)
else:
with open(args.panelfile) as f:
# open a YAML file that describe panel sets
import yaml
Panel.upload(yaml.load(f))
if not args.markerfile:
Marker.upload(params.default_markers)
else:
raise NotImplementedError()
panel = Panel.get_panel(args.panel)
fsa_list = []
index = 1
# prepare caching
if args.use_cache:
if not os.path.exists('.fatools_caches/channels'):
os.makedirs('.fatools_caches/channels')
if args.file:
for fsa_filename in args.file.split(','):
fsa_filename = fsa_filename.strip()
if args.indir != "":
filename = args.indir + "/" + fsa_filename
else:
filename = fsa_filename
fsa = FSA.from_file(filename,
panel,
_params,
cache=not args.no_cache)
# yield (fsa, str(i))
fsa_list.append((fsa, str(index)))
index += 1
elif args.infile:
with open(args.infile) as f:
buf, delim = detect_buffer(f.read())
inrows = csv.DictReader(StringIO(buf), delimiter=delim)
line = 1
index = 1
for r in inrows:
line += 1
fsa_filename = r['FILENAME'].strip()
if fsa_filename.startswith('#'):
continue
if r.get('OPTIONS', None):
options = tokenize(r['OPTIONS'])
else:
options = None
panel_code = r.get('PANEL', None) or args.panel
panel = Panel.get_panel(panel_code)
fsa = FSA.from_file(fsa_filename,
panel,
_params,
options,
cache=not args.no_cache)
if 'SAMPLE' in inrows.fieldnames:
# yield (fsa, r['SAMPLE'])
fsa_list.append((fsa, r['SAMPLE']))
else:
# yield (fsa, str(index))
fsa_list.append((fsa, str(index)))
index += 1
elif args.indir:
import glob
for fsa_filename in sorted(glob.glob(args.indir + "/*.fsa")):
fsa_filename = fsa_filename.strip()
fsa = FSA.from_file(fsa_filename,
panel,
_params,
cache=not args.no_cache)
# yield (fsa, str(i))
fsa_list.append((fsa, str(index)))
index += 1
return fsa_list