Mercurial > repos > miller-lab > genome_diversity

annotate rank_terms.py @ 24:248b06e86022

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Added gd_genotype datatype. Modified tools to support new datatype.
author Richard Burhans <burhans@bx.psu.edu>
date Tue, 28 May 2013 16:24:19 -0400
parents 95a05c1ef5d5
children 8997f2ca8c7a
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
24
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
1 #!/usr/bin/env python
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
2 # -*- coding: utf-8 -*-
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
3 #
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
4 # GOFisher.py
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
5 #
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
6 # Copyright 2013 Oscar Reina <oscar@niska.bx.psu.edu>
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
7 #
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
8 # This program is free software; you can redistribute it and/or modify
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
9 # it under the terms of the GNU General Public License as published by
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
10 # the Free Software Foundation; either version 2 of the License, or
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
11 # (at your option) any later version.
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
12 #
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
13 # This program is distributed in the hope that it will be useful,
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
16 # GNU General Public License for more details.
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
17 #
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
18 # You should have received a copy of the GNU General Public License
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
19 # along with this program; if not, write to the Free Software
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
20 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
21 # MA 02110-1301, USA.
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
22
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
23 import argparse
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
24 import os
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
25 import sys
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
26 from fisher import pvalue
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
27 from decimal import Decimal,getcontext
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
28
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
29 def rtrnGOcENSEMBLc(inExtnddfile,columnENSEMBLTExtndd,columnGOExtndd):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
30 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
31 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
32 dGOTENSEMBLT={}
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
33 for eachl in open(inExtnddfile,'r'):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
34 if eachl.strip():
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
35 ENSEMBLT=eachl.splitlines()[0].split('\t')[columnENSEMBLTExtndd]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
36 GOTs=set(eachl.splitlines()[0].split('\t')[columnGOExtndd].split('.'))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
37 GOTs=GOTs.difference(set(['','U','N']))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
38 for GOT in GOTs:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
39 try:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
40 dGOTENSEMBLT[GOT].add(ENSEMBLT)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
41 except:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
42 dGOTENSEMBLT[GOT]=set([ENSEMBLT])
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
43 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
44 ##dGOTENSEMBLT.pop('')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
45 ENSEMBLTGinGO=set.union(*dGOTENSEMBLT.values())
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
46 return dGOTENSEMBLT,ENSEMBLTGinGO
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
47
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
48 def rtrnENSEMBLcSAPs(inSAPsfile,columnENSEMBLT,ENSEMBLTGinGO):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
49 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
50 returns a set of the ENSEMBLT codes present in the input list and
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
51 in the GO file
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
52 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
53 sENSEMBLTSAPsinGO=set()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
54 for eachl in open(inSAPsfile,'r'):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
55 ENSEMBLT=eachl.splitlines()[0].split('\t')[columnENSEMBLT]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
56 if ENSEMBLT in ENSEMBLTGinGO:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
57 sENSEMBLTSAPsinGO.add(ENSEMBLT)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
58 return sENSEMBLTSAPsinGO
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
59
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
60 def rtrnCounts(dGOTENSEMBLT,ENSEMBLTGinGO,sENSEMBLTSAPsinGO):
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
61 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
62 returns a list of the ENSEMBLT codes present in the input list and
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
63 in the GO file. The terms in this list are: 'Go Term','# Genes in
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
64 the GO Term','# Genes in the list and in the GO Term','Enrichement
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
65 of the GO Term for genes in the input list','Genes in the input list
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
66 present in the GO term'
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
67 """
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
68 getcontext().prec=2#set 2 decimal places
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
69 SAPs_all=len(sENSEMBLTSAPsinGO)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
70 NoSAPs_all=len(ENSEMBLTGinGO)-SAPs_all
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
71 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
72 lp=len(dGOTENSEMBLT)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
73 cnt=0
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
74 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
75 ltfreqs=[]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
76 for echGOT in dGOTENSEMBLT:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
77 cnt+=1
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
78 ##print 'Running "%s", %s out of %s'%(echGOT,cnt,lp)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
79 CntGO_All=len(dGOTENSEMBLT[echGOT])
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
80 SAPs_GO=len(dGOTENSEMBLT[echGOT].intersection(sENSEMBLTSAPsinGO))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
81 NoSAPs_GO=CntGO_All-SAPs_GO
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
82 pval=pvalue(SAPs_GO,NoSAPs_GO,SAPs_all,NoSAPs_all)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
83 #~ outl.append('\t'.join([str(x) for x in(str(pval.two_tail),CntGO_All,SAPs_GO,'.'.join(sorted(dGOTENSEMBLT[echGOT].intersection(sENSEMBLTSAPsinGO))),echGOT)]))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
84 ltfreqs.append([(SAPs_GO/Decimal(CntGO_All)),SAPs_GO,Decimal(str(pval.two_tail))*1,echGOT])
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
85 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
86 ltfreqs.sort()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
87 ltfreqs.reverse()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
88 outl=[]
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
89 cper,crank=Decimal('2'),0
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
90 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
91 for perc,cnt_go,pval,goTerm in ltfreqs:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
92 if perc<cper:
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
93 crank+=1
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
94 cper=perc
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
95 outl.append('\t'.join([str(cnt_go),str(perc),str(crank),str(pval),goTerm]))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
96 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
97 return outl
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
98
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
99
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
100 def main():
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
101 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
102 parser = argparse.ArgumentParser(description='Returns the count of genes in GO categories and their statistical overrrepresentation, from a list of genes and an extended file (i.e. plane text with ENSEMBLT and GO terms).')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
103 parser.add_argument('--input',metavar='input TXT file',type=str,help='the input file with the table in txt format.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
104 parser.add_argument('--inExtnddfile',metavar='input TXT file',type=str,help='the input file with the extended table in txt format.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
105 parser.add_argument('--output',metavar='output TXT file',type=str,help='the output file with the table in txt format.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
106 parser.add_argument('--columnENSEMBLT',metavar='column number',type=int,help='column with the ENSEMBL transcript code in the input file.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
107 parser.add_argument('--columnENSEMBLTExtndd',metavar='column number',type=int,help='column with the ENSEMBL transcript code in the extended file.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
108 parser.add_argument('--columnGOExtndd',metavar='column number',type=int,help='column with the GO terms in the extended file.',required=True)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
109
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
110 args = parser.parse_args()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
111
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
112 inSAPsfile = args.input
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
113 inExtnddfile = args.inExtnddfile
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
114 saleGOPCount = args.output
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
115 columnENSEMBLT = args.columnENSEMBLT
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
116 columnENSEMBLTExtndd = args.columnENSEMBLTExtndd
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
117 columnGOExtndd = args.columnGOExtndd
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
118
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
119 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
120 dGOTENSEMBLT,ENSEMBLTGinGO=rtrnGOcENSEMBLc(inExtnddfile,columnENSEMBLTExtndd,columnGOExtndd)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
121 sENSEMBLTSAPsinGO=rtrnENSEMBLcSAPs(inSAPsfile,columnENSEMBLT,ENSEMBLTGinGO)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
122 outl=rtrnCounts(dGOTENSEMBLT,ENSEMBLTGinGO,sENSEMBLTSAPsinGO)
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
123 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
124 saleGOPCount=open(saleGOPCount,'w')
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
125 saleGOPCount.write('\n'.join(outl))
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
126 saleGOPCount.close()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
127 #~
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
128 return 0
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
129
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
130 if __name__ == '__main__':
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
131 main()
248b06e86022 Added gd_genotype datatype. Modified tools to support new datatype.
Richard Burhans <burhans@bx.psu.edu>
parents: 22
diff changeset
132