v2.6

Use different MLST species scheme map versions
Write to stderr number samples FAIL

INNUca.py

@@ -8,9 +8,9 @@
 INNUca.py - INNUENDO quality control of reads, de novo assembly and contigs quality assessment, and possible contamination search
 <https://github.com/B-UMMI/INNUca>
 
-Copyright (C) 2016 Miguel Machado <[email protected]>
+Copyright (C) 2017 Miguel Machado <[email protected]>
 
-Last modified: April 03, 2017
+Last modified: June 21, 2017
 
 This program is free software: you can redistribute it and/or modify
 it under the terms of the GNU General Public License as published by
@@ -70,7 +70,7 @@ def get_trueCoverage_config(skipTrueCoverage, trueConfigFile, speciesExpected, s
 
 
 def main():
-	version = '2.5'
+	version = '2.6'
 	args = utils.parseArguments(version)
 
 	general_start_time = time.time()
@@ -163,9 +163,8 @@ def main():
 	scheme = 'unknown'
 	if not args.skipMLST and not args.skipSPAdes:
 		scheme = mlst.getScheme(args.speciesExpected)
-
-	# Get path to blastn
-	mlst.getBlastPath()
+		# Print path to blastn
+		mlst.getBlastPath()
 
 	# Get trueCoverage_ReMatCh settings
 	trueCoverage_config = get_trueCoverage_config(args.skipTrueCoverage, args.trueConfigFile.name if args.trueConfigFile is not None else None, args.speciesExpected, script_path)
@@ -260,6 +259,9 @@ def main():
 	# Check whether INNUca.py run at least one sample successfully
 	if number_samples_successfully == 0:
 		sys.exit('No samples run successfully!')
+	else:
+		if number_samples_pass < number_samples_successfully:
+			sys.stderr.write('{fail_samples} samples FAIL INNUca.py analysis'.format(fail_samples=(number_samples_successfully - number_samples_pass)))
 
 
 def get_sample_args_fastq(fastq_files_list, outdir, pairEnd_filesSeparation_list):

modules/mlst.py

@@ -1,18 +1,38 @@
 import utils
 import os
 from functools import partial
+import sys
+
 
 mlst_timer = partial(utils.timer, name='MLST')
 
 
-def getScheme(species):
-	command = ['which', 'mlst']
-	run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, False)
-	mlst_folder = os.path.abspath(os.path.realpath(stdout.splitlines()[0]))
+def get_species_scheme_map_version(mlst_folder):
+	species_scheme_map_version = 1
+
 	mlst_db_path = os.path.join(os.path.dirname(os.path.dirname(mlst_folder)), 'db', 'species_scheme_map.tab')
+	if not os.path.isfile(mlst_db_path):
+		mlst_db_path = os.path.join(os.path.dirname(os.path.dirname(mlst_folder)), 'db', 'scheme_species_map.tab')
+		if not os.path.isfile(mlst_db_path):
+			sys.exit('ERROR: species_scheme_map not found. Contact the developers. In the meantime try running INNUca with --skipMLST option')
+		else:
+			species_scheme_map_version = 2
+	return mlst_db_path, species_scheme_map_version
+
+
+def set_species_scheme_map_variables(list_values, species_scheme_map_version):
+	if species_scheme_map_version == 1:
+		val_genus = list_values[0]
+		val_species = list_values[1]
+		val_scheme = list_values[2]
+	elif species_scheme_map_version == 2:
+		val_genus = list_values[1]
+		val_species = list_values[2]
+		val_scheme = list_values[0]
+	return val_genus, val_species, val_scheme
 
-	species = species.lower().split(' ')
 
+def parse_species_scheme_map(species_splited, mlst_db_path, species_scheme_map_version):
 	scheme = 'unknown'
 
 	with open(mlst_db_path, 'rtU') as reader:
@@ -23,15 +43,27 @@ def getScheme(species):
 				if not line.startswith('#'):
 					line = line.lower().split('\t')
 					line = [line[i].split(' ')[0] for i in range(0, len(line))]
-					if line[0] == species[0]:
-						if line[1] == '':
-							genus_mlst_scheme = line[2]
-						elif line[1] == species[1]:
-							scheme = line[2]
+					val_genus, val_species, val_scheme = set_species_scheme_map_variables(line, species_scheme_map_version)
+					if val_genus == species_splited[0]:
+						if val_species == '':
+							genus_mlst_scheme = val_scheme
+						elif val_species == species_splited[1]:
+							scheme = val_scheme
 		if scheme == 'unknown' and genus_mlst_scheme is not None:
 			scheme = genus_mlst_scheme
+	return scheme
+
+
+def getScheme(species):
+	command = ['which', 'mlst']
+	run_successfully, stdout, stderr = utils.runCommandPopenCommunicate(command, False, None, False)
+	mlst_folder = os.path.abspath(os.path.realpath(stdout.splitlines()[0]))
+
+	mlst_db_path, species_scheme_map_new = get_species_scheme_map_version(mlst_folder)
+
+	scheme = parse_species_scheme_map(species.lower().split(' '), mlst_db_path, species_scheme_map_new)
 
-	print '\n' + 'MLST scheme found for ' + ' '.join(species) + ': ' + scheme
+	print '\n' + 'MLST scheme found for {species}: {scheme}'.format(species=species, scheme=scheme)
 
 	return scheme