help='USEARCH8 EXE')
parser.add_argument('--debug',
action='store_true',
help='Remove Intermediate Files')
args = parser.parse_args()
if not args.out:
#get base name of files
base = args.otu_table.split(".otu_table")
base = base[0]
else:
base = args.out
#remove logfile if exists
log_name = base + '.ufits-filter.log'
ufitslib.removefile(log_name)
ufitslib.setupLogging(log_name)
FNULL = open(os.devnull, 'w')
cmd_args = " ".join(sys.argv) + '\n'
ufitslib.log.debug(cmd_args)
print "-------------------------------------------------------"
#initialize script, log system info and usearch version
ufitslib.SystemInfo()
#Do a version check
usearch = args.usearch
ufitslib.versionDependencyChecks(usearch)
#check if otu_table is empty
ufitslib.log.info("Loading OTU table: %s" % args.otu_table)
def getAvgLength(input):
AvgLength = []
for title, seq, qual in FastqGeneralIterator(open(input)):
AvgLength.append(len(seq))
Average = sum(AvgLength) / float(len(AvgLength))
Min = min(AvgLength)
Max = max(AvgLength)
a = np.array(AvgLength)
nintyfive = np.percentile(a, 5)
return (Average, Min, Max, int(nintyfive))
#remove logfile if exists
log_name = args.out + '.ufits-dada2.log'
if os.path.isfile(log_name):
ufitslib.removefile(log_name)
ufitslib.setupLogging(log_name)
FNULL = open(os.devnull, 'w')
cmd_args = " ".join(sys.argv)+'\n'
ufitslib.log.debug(cmd_args)
print "-------------------------------------------------------"
#initialize script, log system info and usearch version
ufitslib.SystemInfo()
#Do a version check
usearch = args.usearch
ufitslib.versionDependencyChecks(usearch)
#check dependencies
programs = ['Rscript']
ufitslib.CheckDependencies(programs)
ufitslib.log.error("Blast taxonomy is not compatible with BIOM output, use a different method")
#create OTU phylogeny for downstream processes
ufitslib.log.info("Generating phylogenetic tree")
tree_out = base + '.tree.phy'
cmd = [usearch, '-cluster_agg', args.fasta, '-treeout', tree_out]
ufitslib.runSubprocess(cmd, ufitslib.log)
#print some summary file locations
ufitslib.log.info("Taxonomy finished: %s" % taxFinal)
if args.otu_table and not args.method == 'blast':
ufitslib.log.info("Classic OTU table with taxonomy: %s" % taxTable)
#output final OTU table in Biom v1.0 (i.e. json format if biom installed)
outBiom = base + '.biom'
if ufitslib.which('biom'):
ufitslib.removefile(outBiom)
cmd = ['biom', 'convert', '-i', tmpTable, '-o', outBiom+'.tmp', '--table-type', "OTU table", '--to-json']
ufitslib.runSubprocess(cmd, ufitslib.log)
if args.mapping_file:
cmd = ['biom', 'add-metadata', '-i', outBiom+'.tmp', '-o', outBiom, '--observation-metadata-fp', qiimeTax, '-m', args.mapping_file, '--sc-separated', 'taxonomy']
ufitslib.runSubprocess(cmd, ufitslib.log)
else:
cmd = ['biom', 'add-metadata', '-i', outBiom+'.tmp', '-o', outBiom, '--observation-metadata-fp', qiimeTax, '--sc-separated', 'taxonomy']
ufitslib.runSubprocess(cmd, ufitslib.log)
ufitslib.removefile(outBiom+'.tmp')
ufitslib.log.info("BIOM OTU table created: %s" % outBiom)
else:
ufitslib.log.info("biom program not installed, install via `pip install biom-format`")
ufitslib.log.info("OTUs with taxonomy: %s" % otuTax)
ufitslib.log.info("OTU phylogeny: %s" % tree_out)
parser.add_argument('-o','--out', help='Base output name')
parser.add_argument('--min_reads_otu', default=2, type=int, help='Minimum number of reads per OTU for experiment')
parser.add_argument('-u','--usearch', dest="usearch", default='usearch9', help='USEARCH8 EXE')
parser.add_argument('--debug', action='store_true', help='Remove Intermediate Files')
args=parser.parse_args()
if not args.out:
#get base name of files
base = args.otu_table.split(".otu_table")
base = base[0]
else:
base = args.out
#remove logfile if exists
log_name = base + '.ufits-filter.log'
ufitslib.removefile(log_name)
ufitslib.setupLogging(log_name)
FNULL = open(os.devnull, 'w')
cmd_args = " ".join(sys.argv)+'\n'
ufitslib.log.debug(cmd_args)
print "-------------------------------------------------------"
#initialize script, log system info and usearch version
ufitslib.SystemInfo()
#Do a version check
usearch = args.usearch
ufitslib.versionDependencyChecks(usearch)
#check if otu_table is empty
ufitslib.log.info("Loading OTU table: %s" % args.otu_table)
