fix bug in DMRfind due to recent modification on filter-allc

Author: yupenghe

methylpy/DMRfind.py

@@ -7,7 +7,7 @@
 from multiprocessing import Pool
 from methylpy.utilities import print_checkpoint,expand_nucleotide_code
 from methylpy.utilities import open_allc_file,split_files_by_position
-from methylpy.utilities import filter_allc_file,index_allc_file_batch
+from methylpy.utilities import filter_allc_files,index_allc_file_batch
 from methylpy.utilities import read_allc_index,remove_allc_index
 from scipy.stats import scoreatpercentile
 import subprocess
@@ -129,35 +129,19 @@ def DMRfind(allc_files, samples,
 
 
     # filter allc file
-    if num_procs > 1:
-        pool = Pool(min(num_procs,len(samples)))
-    else:
-        pool = False
     query_allc_files = []
     for allc_file,sample in zip(allc_files,samples):
         query_allc_files.append(output_prefix+"_filtered_allc_"+sample+".tsv")
-        if pool:
-            pool.apply_async(filter_allc_file,
-                             (),
-                             {"allc_file":allc_file,
-                              "output_file":output_prefix+"_filtered_allc_"+sample+".tsv",
-                              "mc_type":mc_type,
-                              "chroms":chroms,
-                              "compress_output":False,
-                              "min_cov":min_cov,
-                              "buffer_line_number":buffer_line_number}
-            )
-        else:
-            filter_allc_file(allc_file=allc_file,
-                             output_file=output_prefix+"_filtered_allc_"+sample+".tsv",
-                             mc_type=mc_type,
-                             chroms=chroms,
-                             compress_output=False,
-                             min_cov=min_cov,
-                             buffer_line_number=buffer_line_number)
-    if pool:
-        pool.close()
-        pool.join()
+
+    filter_allc_files(allc_files=allc_files,
+                      output_files=[output_prefix+"_filtered_allc_"+sample+".tsv" \
+                                    for sample in samples],
+                      num_procs=num_procs,
+                      mc_type=mc_type,
+                      chroms=chroms,
+                      compress_output=False,
+                      min_cov=min_cov,
+                      buffer_line_number=buffer_line_number)
 
     # scan allc file to set up a table for fast look-up of lines belong
     # to different chromosomes

methylpy/call_mc_se.py

@@ -855,12 +855,18 @@ def run_bowtie(current_library,library_read_files,
     if num_procs > 1:
         pool = multiprocessing.Pool(num_procs)
         for file_num in range(0,num_procs):
-            pool.apply_async(subprocess.check_call,(shlex.split("env LC_COLLATE=C sort" + sort_option + " -t '\t' -k 1 -o "+prefix+"_sorted_"+str(file_num)+" "+prefix+"_sorted_"+str(file_num)),))
+            pool.apply_async(subprocess.check_call,
+                             (shlex.split(
+                                 "env LC_COLLATE=C sort"+sort_option+ \
+                                 " -t '\t' -k 1 -o "+prefix+"_sorted_"+str(file_num)+ \
+                                 " "+prefix+"_sorted_"+str(file_num)),))
         pool.close()
         pool.join()
     else:
         for file_num in range(0,num_procs):
-            subprocess.check_call(shlex.split("env LC_COLLATE=C sort" + sort_option + " -t '\t' -k 1 -o "+prefix+"_sorted_"+str(file_num)+" "+prefix+"_sorted_"+str(file_num)))
+            subprocess.check_call(shlex.split(
+                "env LC_COLLATE=C sort"+sort_option + " -t '\t' -k 1 -o "+ \
+                prefix+"_sorted_"+str(file_num)+" "+prefix+"_sorted_"+str(file_num)))
 
     print_checkpoint("Finding multimappers")
 

methylpy/parser.py

@@ -245,16 +245,16 @@ def parse_args():
                                 num_procs=args.num_procs,
                                 no_reindex=args.no_reindex)
      elif args.command == "filter-allc":
-          from methylpy.utilities import filter_allc_file
-          filter_allc_file(allc_files=args.allc_files,
-                           output_files=args.output_files,
-                           num_procs=args.num_procs,
-                           mc_type=args.mc_type,
-                           chroms=args.chroms,
-                           compress_output=args.compress_output,
-                           max_mismatch=args.max_mismatch,
-                           max_mismatch_frac=args.max_mismatch_frac,
-                           min_cov=args.min_cov)
+          from methylpy.utilities import filter_allc_files
+          filter_allc_files(allc_files=args.allc_files,
+                            output_files=args.output_files,
+                            num_procs=args.num_procs,
+                            mc_type=args.mc_type,
+                            chroms=args.chroms,
+                            compress_output=args.compress_output,
+                            max_mismatch=args.max_mismatch,
+                            max_mismatch_frac=args.max_mismatch_frac,
+                            min_cov=args.min_cov)
 
      elif args.command == "allc-to-bigwig":
           from methylpy.utilities import convert_allc_to_bigwig

methylpy/utilities.py

@@ -191,16 +191,16 @@ def convert_allc_to_bigwig(input_allc_file,
     subprocess.check_call(shlex.split("rm "+output_file+".wig "+output_file+".chrom_size"))
 
 
-def filter_allc_file(allc_files,
-                     output_files,
-                     num_procs=1,
-                     mc_type=None,
-                     chroms = None,
-                     compress_output=False,
-                     min_cov=0,
-                     max_mismatch=None,
-                     max_mismatch_frac=None,
-                     buffer_line_number=100000):
+def filter_allc_files(allc_files,
+                      output_files,
+                      num_procs=1,
+                      mc_type=None,
+                      chroms = None,
+                      compress_output=False,
+                      min_cov=0,
+                      max_mismatch=None,
+                      max_mismatch_frac=None,
+                      buffer_line_number=100000):
 
     # User input checks
     if not isinstance(allc_files, list):

setup.py

@@ -2,7 +2,7 @@
 
 setup(
     name='methylpy',
-    version='1.1.9',
+    version='1.2.1',
     author='Yupeng He',
     author_email='yupeng.he.bioinfo@gmail.com',
     packages=['methylpy'],