change Epinano_Variants.py! no sliding windown analysis

tsjzz · Oct 5, 2020 · 0535bdf · 0535bdf
1 parent 3accdcb
commit 0535bdf
Showing 1 changed file with 21 additions and 13 deletions.
diff --git a/Epinano_Variants.py b/Epinano_Variants.py
@@ -139,15 +139,20 @@ def df_groupy(df, out):
 
 #~~~~~~~~~~~~~~~~~~~~~~~ main () ~~~~~~~~~~~~~~~~~~~~~~~
 parser = argparse.ArgumentParser()
-#parser.add_argument ('-r','--reads',help='fastq(a) reads input')
-parser.add_argument ('-R','--reference', help='samtools faidx indexed reference file')
-parser.add_argument ('-b', '--bam', type=str, help='bam file; if given; no need to offer reads file; mapping will be skipped')
+required_args = parser.add_argument_group ('Required Arguments')
+required_args.add_argument ('-R','--reference', help='samtools faidx indexed reference file')
+required_args.add_argument ('-b', '--bam', type=str, help='bam file; if given; no need to offer reads file; mapping will be skipped')
+required_args.add_argument ('-s', '--sam2tsv',type=str, default='',help='/path/to/sam2tsv.jar; needed unless a sam2tsv.jar produced file is already given')
 parser.add_argument ('-f','--file', type=str, help='tsv file generated by sam2tsv.jar; if given, reads mapping and sam2tsv conversion will be skipped')
 parser.add_argument ('-t', '--threads', type=int, default=4,  help='number of threads') 
-parser.add_argument ('-s', '--sam2tsv',type=str, default='',help='/path/to/sam2tsv.jar; needed unless a sam2tsv.jar produced file is already given')
 parser.add_argument ('-T', '--type', type=str, default="t" ,help="reference types, which is either g(enome) or t(ranscriptome);")
+parser.add_argument ('-w', '--window', type=int, default=0, help= """specify a windown/kmer size to re-organize per site variants on kmer/window basis;
+								By default, this is not performed;
+								 It is recommended to do it seperately with EpiNano/misc/Slide_Variants.py """)
+
 parser.add_argument ('-p','--per_read_variants', action='store_true', help='compute per reads variants statistics')
 args=parser.parse_args()
+
 #~~~~~~~~~~~~~~~~~~~~~~~ prepare for analysis ~~~~~~~~~~~~~~ 
 tsv_gen = None  # generator 
 prefix = '' 
@@ -161,7 +166,7 @@ def _tsv_gen ():
 				exit()
 			else:
 				if not file_exist (args.sam2tsv):
-					sys.stderr.write (" can  not find {} java program\n".format(args.sam2tsv))
+					sys.stderr.write ("Please offer correctly path to sam2tsv.jar\n".format(args.sam2tsv))
 					exit()
 				if not os.path.exists (bam_file+'.bai'):
 					os.system ('samtools index ' + bam_file + '.bai')
@@ -265,15 +270,18 @@ def _tsv_gen ():
 	tmp_files = glob.glob("{}/small*".format(tmp_dir))
 	pool.map(_rm,  tmp_files)
 	shutil.rmtree(tmp_dir)
-
+#print ("per site variants frequencies table has been generated", file = sys.stderr)
 #3 sliding window per site variants --> for making predicitons 
-if len (var_files) == 1:
-	slide_per_site_var(var_files[0])
-elif len (var_files) == 2:
-	pool = mp.Pool(2)
-	pool.map(slide_per_site_var, var_files)
-	pool.close(); pool.join()
-
+if args.window:
+	print (args.window,'mer')
+	kmer = args.window 
+	print ("slice per site variants frequencies table on {}mer basis".format (kmer), file=sys.stderr)
+	if len (var_files) == 1:
+		slide_per_site_var(var_files[0], kmer)
+	elif len (var_files) == 2:
+		pool = mp.Pool(2)
+		pool.starmap(slide_per_site_var, (var_files,kmer))
+		pool.close(); pool.join()
 
 # per read variants 
 if args.per_read_variants: