def validate_input_create(main_dir, args):
'''
Accepts:
- main_dir
- docopt args
Returns:
- title
- fasta_f
- fasta_type
- cov_libs
- hit_libs
- nodesDB_f
- taxrules
- out_f
'''
ASSEMBLY_TYPES = [None, 'spades', 'soap', 'abyss', 'velvet']
fasta_f = args['--infile']
fasta_type = args['--type']
sam_fs = args['--sam']
bam_fs = args['--bam']
cov_fs = args['--cov']
cas_fs = args['--cas']
hit_fs = args['--taxfile']
out_f = args['--out']
if (out_f):
out_f = "%s.%s" % (os.path.basename(out_f), "BlobDB.json")
else:
out_f = "%s" % ("BlobDB.json")
nodesDB_f = args['--db']
names_f = args['--names']
nodes_f = args['--nodes']
taxrules = args['--taxrule']
title = args['--title'] if (args['--title']) else out_f
# Do files exist ?
files = [
x for x in list([fasta_f] + sam_fs + bam_fs + cov_fs + cas_fs +
[names_f] + [nodes_f] + hit_fs) if x is not None
]
for f in files:
if not os.path.isfile(f):
BtLog.error('0', f)
# Is taxonomy provided?
if nodesDB_f == "data/nodesDB.txt":
nodesDB_f = os.path.join(main_dir, nodesDB_f)
if not os.path.isfile(nodesDB_f) and not ((names_f) and (nodes_f)):
BtLog.error('3')
if not (hit_fs):
BtLog.error('18')
# can FASTA parser deal with assemblies
if not fasta_type in ASSEMBLY_TYPES:
BtLog.error('2', ",".join(ASSEMBLY_TYPES[1:]))
# Is coverage provided?
if not (fasta_type
) and not bam_fs and not sam_fs and not cov_fs and not cas_fs:
BtLog.error('1')
cov_libs = [bt.CovLibObj('bam' + str(idx), 'bam', lib_f) for idx, lib_f in enumerate(bam_fs)] + \
[bt.CovLibObj('sam' + str(idx), 'sam', lib_f) for idx, lib_f in enumerate(sam_fs)] + \
[bt.CovLibObj('cas' + str(idx), 'cas', lib_f) for idx, lib_f in enumerate(cas_fs)] + \
[bt.CovLibObj('cov' + str(idx), 'cov', lib_f) for idx, lib_f in enumerate(cov_fs)]
hit_libs = [
bt.hitLibObj('tax' + str(idx), 'tax', lib_f)
for idx, lib_f in enumerate(hit_fs)
]
return title, fasta_f, fasta_type, cov_libs, hit_libs, taxrules, nodesDB_f, nodes_f, names_f, out_f
cov_libs = [bt.CovLibObj('bam' + str(idx), 'bam', lib_f) for idx, lib_f in enumerate(bam_fs)] + \
[bt.CovLibObj('sam' + str(idx), 'sam', lib_f) for idx, lib_f in enumerate(sam_fs)] + \
[bt.CovLibObj('cas' + str(idx), 'cas', lib_f) for idx, lib_f in enumerate(cas_fs)] + \
[bt.CovLibObj('cov' + str(idx), 'cov', lib_f) for idx, lib_f in enumerate(cov_fs)]
# Create BlobDB object
blobDb = bt.BlobDb(title)
# Parse FASTA
blobDb.parseFasta(fasta_f, fasta_type)
# Parse coverage
blobDb.parseCovs(cov_libs)
# Parse Tax
hitLibs = [bt.hitLibObj('tax' + str(idx), 'tax', lib_f) for idx, lib_f in enumerate(hit_fs)]
blobDb.parseHits(hitLibs)
# Parse nodesDB
nodesDB, nodesDB_f = BtIO.getNodesDB(nodes=nodes_f, names=names_f, nodesDB=nodesDB_f)
blobDb.nodesDB_f = nodesDB_f
if not os.path.isfile(nodesDB_f):
print BtLog.status_d['5'] % nodesDB_f
BtIO.writeNodesDB(nodesDB, nodesDB_f)
# Computing taxonomy based on taxrules
print BtLog.status_d['6'] % ",".join(taxrules)
blobDb.computeTaxonomy(taxrules, nodesDB)
# Generating BlobDB and writing to file
def validate_input_create(main_dir, args):
'''
Accepts:
- main_dir
- docopt args
Returns:
- title
- fasta_f
- fasta_type
- cov_libs
- hit_libs
- nodesDB_f
- taxrules
- out_f
'''
ASSEMBLY_TYPES = [None, 'spades', 'soap', 'abyss', 'velvet']
fasta_f = args['--infile']
fasta_type = args['--type']
sam_fs = args['--sam']
bam_fs = args['--bam']
cov_fs = args['--cov']
cas_fs = args['--cas']
hit_fs = args['--taxfile']
out_f = args['--out']
if (out_f):
out_f = "%s.%s" % (os.path.basename(out_f), "BlobDB.json")
else:
out_f = "%s" % ("BlobDB.json")
nodesDB_f = args['--db']
names_f = args['--names']
nodes_f = args['--nodes']
taxrules = args['--taxrule']
title = args['--title'] if (args['--title']) else out_f
# Do files exist ?
files = [x for x in list([fasta_f] + sam_fs + bam_fs + cov_fs + cas_fs + [names_f] + [nodes_f] + hit_fs) if x is not None]
for f in files:
if not os.path.isfile(f):
BtLog.error('0', f)
# Is taxonomy provided?
if nodesDB_f == "data/nodesDB.txt":
nodesDB_f = os.path.join(main_dir, nodesDB_f)
if not os.path.isfile(nodesDB_f) and not ((names_f) and (nodes_f)):
BtLog.error('3')
if not (hit_fs):
BtLog.error('18')
# can FASTA parser deal with assemblies
if not fasta_type in ASSEMBLY_TYPES:
BtLog.error('2', ",".join(ASSEMBLY_TYPES[1:]))
# Is coverage provided?
if not (fasta_type) and not bam_fs and not sam_fs and not cov_fs and not cas_fs:
BtLog.error('1')
cov_libs = [bt.CovLibObj('bam' + str(idx), 'bam', lib_f) for idx, lib_f in enumerate(bam_fs)] + \
[bt.CovLibObj('sam' + str(idx), 'sam', lib_f) for idx, lib_f in enumerate(sam_fs)] + \
[bt.CovLibObj('cas' + str(idx), 'cas', lib_f) for idx, lib_f in enumerate(cas_fs)] + \
[bt.CovLibObj('cov' + str(idx), 'cov', lib_f) for idx, lib_f in enumerate(cov_fs)]
hit_libs = [bt.hitLibObj('tax' + str(idx), 'tax', lib_f) for idx, lib_f in enumerate(hit_fs)]
return title, fasta_f, fasta_type, cov_libs, hit_libs, taxrules, nodesDB_f, nodes_f, names_f, out_f