Merge pull request (python3 support) from HyperOdin-master

Changes:

1. all usages of python 2-"file"'s are replaced with python 3-
   "open"'s
2. import six for python 2 and 3 compatibility
3. replace python 2-commands module with python 3-subprocess
   module
4. replace python 2-"print x" with python 3-"print (x)"

bin/CPC2.py

@@ -5,14 +5,13 @@
 import sys
 import os
 import re
-import commands
+import subprocess
 import time
 from optparse import OptionParser,OptionGroup
-
+import six
 import numpy as np
 from Bio.Seq import Seq
 from Bio.SeqUtils import ProtParam
-
 import seqio
 
 def __main():
@@ -85,7 +84,7 @@ def find_longest_in_one(self,myframe,direction,start_codon,stop_codon):
  '''
  while True:
  try: 
- codon,index = triplet_got.next()
+ codon,index = next(triplet_got)
  except StopIteration:
  break 
  if codon in starts and codon not in stops:
@@ -96,7 +95,7 @@ def find_longest_in_one(self,myframe,direction,start_codon,stop_codon):
  end_extension = False
  while True:
  try: 
- codon,index = triplet_got.next()
+ codon,index = next(triplet_got)
  except StopIteration:
  end_extension = True
  integrity = -1
@@ -252,9 +251,12 @@ def calculate_potential(fasta,strand,output_orf,outfile):
  '''
  strinfoAmbiguous = re.compile("X|B|Z|J|U",re.I)
  ptU = re.compile("U",re.I)
- ftmp_feat = file(outfile + ".feat","w")
- ftmp_svm = file(outfile + ".tmp.1","w")
- ftmp_result = file(outfile,"w")
+## merged by Yang Ding on 2019-11-23
+## 1. all python 2-"file"'s are replaced with python 3-"open"'s
+## 2. keep kangyj's check on output_orf
+ ftmp_feat = open(outfile + ".feat","w")
+ ftmp_svm = open(outfile + ".tmp.1","w")
+ ftmp_result = open(outfile,"w")
  if output_orf == 1:
  my_header = ["#ID","transcript_length","peptide_length","Fickett_score","pI","ORF_integrity","ORF_Start","coding_probability","label"]
  else:
@@ -315,7 +317,7 @@ def calculate_potential(fasta,strand,output_orf,outfile):
  cmd = cmd + app_svm_predict + ' -b 1 -q ' + outfile + '.tmp.2 ' + data_dir + 'cpc2.model ' + outfile + '.tmp.out'
  #cmd = cmd + 'awk -vOFS="\\t" \'{if ($1 == 1){print $2,"coding"} else if ($1 == 0){print $2,"noncoding"}}\' ' + outfile + '.tmp.1 > ' + outfile + '.tmp.2 &&'
  #cmd = cmd + 'paste ' + outfile + '.feat ' + outfile + '.tmp.2 >>' + outfile
- (exitstatus, outtext) = commands.getstatusoutput(cmd) 
+ (exitstatus, outtext) = subprocess.getstatusoutput(cmd) 
 
  '''deal with the output'''
  #print outfile + '.tmp.out'
@@ -352,7 +354,7 @@ def calculate_potential(fasta,strand,output_orf,outfile):
  if exitstatus == 0:
  os.system('rm -f ' + outfile + '.tmp.1 ' + outfile + '.tmp.2 ' + outfile + '.tmp.out ' + outfile)
  rm_cmd = "rm -f " + outfile + '.feat'
- commands.getstatusoutput(rm_cmd)
+ subprocess.getstatusoutput(rm_cmd)
  sys.stderr.write("[INFO] Running Done!\n")
  return 0
  else:

bin/compress.py

@@ -11,7 +11,7 @@ def gz_file(fq_file,mode,level=6):
  fq_fp = gzip.open(fq_file,mode+"b",level)
  else:
  sys.stderr.write("[INFO] read file '%s'\n"%fq_file)
- fq_fp = file(fq_file,mode)
+ fq_fp = open(fq_file,mode)
  except:
  sys.stderr.write("Error: Fail to IO file: %s\n"%(fq_file))
  sys.exit(1)

bin/seqio.py

@@ -162,8 +162,8 @@ def gtf_parse(fn,add="chr"):
 
 if __name__ == "__main__":
  a = [[1,10],[17,22],[40,44],[42,47],[46,100],[101,408]]
- print a
- print merge_region(a)
+ print (a)
+ print (merge_region(a))