Mercurial > repos > miller-lab > genome_diversity

annotate make_phylip.py @ 37:884ccb07885b

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Fixed output_id
author Richard Burhans <burhans@bx.psu.edu>
date Wed, 20 Nov 2013 16:43:43 -0500
parents 51cd0307fb70
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
1 #!/usr/bin/env python
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
2 # -*- coding: utf-8 -*-
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
3 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
4 # mkFastas.py
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
5 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
6 # Copyright 2013 Oscar Reina <oscar@niska.bx.psu.edu>
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
7 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
8 # This program is free software; you can redistribute it and/or modify
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
9 # it under the terms of the GNU General Public License as published by
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
10 # the Free Software Foundation; either version 2 of the License, or
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
11 # (at your option) any later version.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
12 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
13 # This program is distributed in the hope that it will be useful,
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
14 # but WITHOUT ANY WARRANTY; without even the implied warranty of
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
15 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
16 # GNU General Public License for more details.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
17 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
18 # You should have received a copy of the GNU General Public License
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
19 # along with this program; if not, write to the Free Software
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
20 # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston,
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
21 # MA 02110-1301, USA.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
22
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
23 import argparse
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
24 import glob
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
25 import os
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26 import shutil
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
27 from Population import Population
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29 def revseq(seq):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30 seq=list(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31 seq.reverse()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 seq=''.join(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 return seq
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
35 def revComp(allPop):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 dAllCompAll={'A':'T','T':'A','C':'G','G':'C','N':'N','M':'K','K':'M','R':'Y','Y':'R','W':'W','S':'S'}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 allPopsComp=dAllCompAll[allPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38 return allPopsComp
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40 def rtrnCons(ntA,ntB):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41 srtdPairs=''.join(sorted([ntA,ntB]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 dpairsCons={'AC':'M', 'AG':'R', 'AT':'W', 'CG':'S', 'CT':'Y', 'GT':'K', 'AN':'A', 'CN':'C', 'GN':'G', 'NT':'T'}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43 cons=dpairsCons[srtdPairs]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44 return cons
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
45
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46 def rtrnFxdChrPos(inSNPf,dPopsinSNPfPos,pxchrx,pxpos,pxntA,pxntB,fulldChrdPosdPopsAlllsInit=False,cvrgTreshold=False,indvlsPrctTrshld=False):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
49 dChrdPosdPopsAlllsInit={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 seqref=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
51 for eachl in open(inSNPf,'r'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
52 if eachl.strip() and eachl[0]!='#':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
53 fllInfoSplt=eachl.splitlines()[0].split('\t')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
54 chrx=fllInfoSplt[pxchrx]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 pos=int(fllInfoSplt[pxpos])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56 ntA=fllInfoSplt[pxntA]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 ntB=fllInfoSplt[pxntB]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 seqref.append([pos,ntA])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 dPopsAllls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60 if fulldChrdPosdPopsAlllsInit:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62 cntIndv=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
65 fulldPopsAllls=fulldChrdPosdPopsAlllsInit[chrx][pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
66 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67 fulldPopsAllls=dict([(echPop,ntA) for echPop in dPopsinSNPfPos])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
68 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
69 for eachPop in dPopsinSNPfPos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
70 clmnCvrg=dPopsinSNPfPos[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
71 if clmnCvrg:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
72 eachPopCvrg=int(fllInfoSplt[clmnCvrg])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
73 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
74 #~ eachPopCvrg=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
75 eachPopCvrg=cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
76 if eachPopCvrg>=cvrgTreshold:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
77 dPopsAllls[eachPop]=fulldPopsAllls[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
78 cntIndv+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
79 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
80 dPopsAllls[eachPop]='N'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
81 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
82 if indvlsPrctTrshld>(cntIndv/float(len(dPopsinSNPfPos))):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
83 dPopsAllls=dict([(echPop,'N') for echPop in dPopsinSNPfPos])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
84 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
85 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
86 for eachPop in dPopsinSNPfPos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
87 if dPopsinSNPfPos[eachPop]:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
88 eachPopAll=int(fllInfoSplt[dPopsinSNPfPos[eachPop]])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
89 if eachPopAll==0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
90 dPopsAllls[eachPop]=ntB
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
91 elif eachPopAll==2:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
92 dPopsAllls[eachPop]=ntA
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
93 elif eachPopAll==1:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
94 dPopsAllls[eachPop]=rtrnCons(ntA,ntB)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
95 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
96 dPopsAllls[eachPop]='N'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
97 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
98 dPopsAllls[eachPop]=ntA
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
99 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
100 dChrdPosdPopsAlllsInit[chrx][pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
101 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
102 dChrdPosdPopsAlllsInit[chrx]={pos:dPopsAllls}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
103 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
104 seqref.sort()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
105 startExs=[seqref[0][0]]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
106 endExs=[seqref[-1][0]+1]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
107 seqref=''.join(x[1] for x in seqref)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
108 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
109 return dChrdPosdPopsAlllsInit,seqref,chrx,startExs,endExs
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
110
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
111
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
112 def rtrndENSEMBLTseq(inCDSfile,inUCSCfile,fchrClmn,txStartClmn,txEndClmn,strandClmn,geneNameClmn,startExsClmn,endExsClmn,cdsStartClmn,cdsEndClmn):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
113 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
114 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
115 dENSEMBLTchrxStEndEx={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
116 dChrdStrtEndENSEMBLT={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
117 for eachl in open(inUCSCfile,'r'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
118 if eachl.strip():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
119 rvrse=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
120 allVls=eachl.split('\t')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
121 txStart=allVls[txStartClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
122 txEnd=allVls[txEndClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
123 ENSEMBLT=allVls[geneNameClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
124 strand=allVls[strandClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
125 chrx=allVls[fchrClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
126 if cdsStartClmn and cdsEndClmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
127 cdsStart=allVls[cdsStartClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
128 cdsEnd=allVls[cdsEndClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
129 if startExsClmn and endExsClmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
130 startExs=allVls[startExsClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
131 endExs=allVls[endExsClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
132 if strand=='-':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
133 rvrse=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
134 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
135 dChrdStrtEndENSEMBLT[chrx][int(txStart),int(txEnd)]=ENSEMBLT
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
136 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
137 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
138 dChrdStrtEndENSEMBLT[chrx]={(int(txStart),int(txEnd)):ENSEMBLT}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
139 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
140 dChrdStrtEndENSEMBLT={chrx:{(int(txStart),int(txEnd)):ENSEMBLT}}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
141 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
142 if cdsStartClmn and cdsEndClmn and startExsClmn and endExsClmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
143 startExs,endExs=rtrnExnStarEndCorrc(startExs,endExs,cdsStart,cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
144 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
145 startExs,endExs=[int(txStart)],[int(txEnd)]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
146 dENSEMBLTchrxStEndEx[ENSEMBLT]=(chrx,startExs,endExs,rvrse)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
147 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
148 dENSEMBLTseq={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
149 ENSEMBLTseqs=[(x.splitlines()[0],''.join(x.splitlines()[1:])) for x in open(inCDSfile).read().split('>') if x.strip()]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
150 for ENSEMBLT,seq in ENSEMBLTseqs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
151 dENSEMBLTseq[ENSEMBLT]=seq
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
152 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
153 dENSEMBLTseqChrStEnEx={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
154 for ENSEMBLT in dENSEMBLTchrxStEndEx:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
155 chrx,startExs,endExs,rvrse=dENSEMBLTchrxStEndEx[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
156 addEseqChrStEnEx=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
157 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
158 seq=dENSEMBLTseq[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
159 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
160 seq=revseq(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
161 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
162 addEseqChrStEnEx=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
163 if addEseqChrStEnEx:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
164 dENSEMBLTseqChrStEnEx[ENSEMBLT]=(seq,chrx,startExs,endExs,rvrse)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
165 return dENSEMBLTseqChrStEnEx,dChrdStrtEndENSEMBLT
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
166
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
167
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
168 def rtrnFxdChrPosinCodReg(dChrdStrtEndENSEMBLT,dChrdPosdPopsAlllsInit):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
169 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
170 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
171 dENSEMBLTChrPosdAlls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
172 dChrPosdPopsAllls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
173 todel=set(dChrdPosdPopsAlllsInit.keys()).difference(set(dChrdStrtEndENSEMBLT.keys()))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
174 for x in todel:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
175 x=dChrdPosdPopsAlllsInit.pop(x)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
176 #---
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
177 while len(dChrdPosdPopsAlllsInit)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
178 chrx=dChrdPosdPopsAlllsInit.keys()[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
179 dStrtEndENSEMBLT=dChrdStrtEndENSEMBLT.pop(chrx)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
180 dPosdPopsAllls=dChrdPosdPopsAlllsInit.pop(chrx)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
181 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
182 srtdStrtEndENSEMBLT=sorted(dStrtEndENSEMBLT.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
183 srtdPosdPopsAllls=sorted(dPosdPopsAllls.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
184 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
185 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
186 strt,end=srtdStrtEndENSEMBLT.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
187 ENSEMBLT=dStrtEndENSEMBLT[strt,end]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
188 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
189 keePloop=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
190 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
191 while keePloop:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
192 if strt<=pos<=end:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
193 for tmpstrt,tmpend in [(strt,end)]+srtdStrtEndENSEMBLT:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
194 if tmpstrt<=pos<=tmpend:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
195 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
196 dChrPosdPopsAllls[chrx,pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
197 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
198 dENSEMBLTChrPosdAlls[ENSEMBLT][chrx,pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
199 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
200 dENSEMBLTChrPosdAlls[ENSEMBLT]={(chrx,pos):dPopsAllls}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
201 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
202 continue
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
203 if len(srtdPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
204 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
205 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
206 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
207 keePloop=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
208 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
209 elif pos<=strt:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
210 if len(srtdPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
211 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
212 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
213 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
214 keePloop=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
215 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
216 if len(srtdStrtEndENSEMBLT)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
217 strt,end=srtdStrtEndENSEMBLT.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
218 ENSEMBLT=dStrtEndENSEMBLT[strt,end]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
219 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
220 keePloop=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
221 return dENSEMBLTChrPosdAlls,dChrPosdPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
222
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
223 def rtrnExnStarEndCorrc(startExs,endExs,cdsStart,cdsEnd):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
224 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
225 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
226 cdsStart,cdsEnd=int(cdsStart),int(cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
227 crrctdstartExs=set([int(x) for x in startExs.split(',') if x.strip()])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
228 crrctdendExs=set([int(x) for x in endExs.split(',') if x.strip()])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
229 crrctdstartExs.add(cdsStart)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
230 crrctdendExs.add(cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
231 sStartDel=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
232 sEndDel=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
233 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
234 for echvl in crrctdstartExs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
235 if echvl<cdsStart or echvl>cdsEnd:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
236 sStartDel.add(echvl)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
237 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
238 for echvl in crrctdendExs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
239 if echvl<cdsStart or echvl>cdsEnd:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
240 sEndDel.add(echvl)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
241 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
242 return sorted(crrctdstartExs.difference(sStartDel)),sorted(crrctdendExs.difference(sEndDel))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
243
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
244 def rtrndPopsFasta(seq,chrx,startExs,endExs,rvrse,dChrPosdPopsAllls,ENSEMBLT):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
245 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
246 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
247 exnIntrvl=zip(startExs,endExs)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
248 CDSinitPos=exnIntrvl[0][0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
249 dexnIntrvlSeq={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
250 for exStart,exEnd in exnIntrvl:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
251 lenEx=exEnd-exStart
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
252 dexnIntrvlSeq[exStart,exEnd]=seq[:lenEx]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
253 seq=seq[lenEx:]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
254
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
255 ldexnIntrvlSeq=len(dexnIntrvlSeq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
256 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
257 dPopsFasta={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
258 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
259 strePos=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
260 dStrePosAbsPos={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
261 tmpAcmltdPos=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
262 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
263 exStart,exEnd=sorted(dexnIntrvlSeq.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
264 seq=dexnIntrvlSeq.pop((exStart,exEnd))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
265 chrx,pos=sorted(dChrPosdPopsAllls.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
266 dPopsAllls=dChrPosdPopsAllls.pop((chrx,pos))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
267 tmpdPopsFasta=dict([(x,list(seq)) for x in dPopsAllls])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
268 cntExns=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
269 while True:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
270 if exStart<=pos<=exEnd-1:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
271 relPos=tmpAcmltdPos+pos-exStart
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
272 strePos.add(relPos)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
273 dStrePosAbsPos[relPos]=pos
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
274 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
275 allPop=dPopsAllls[echPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
276 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
277 allPop=revComp(allPop)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
278 tmpdPopsFasta[echPop][pos-exStart]=allPop
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
279 if len(dChrPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
280 chrx,pos=sorted(dChrPosdPopsAllls.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
281 dPopsAllls=dChrPosdPopsAllls.pop((chrx,pos))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
282 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
283 pos=endExs[-1]+100#max pos of exns
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
284 elif pos<exStart:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
285 if len(dChrPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
286 chrx,pos=sorted(dChrPosdPopsAllls.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
287 dPopsAllls=dChrPosdPopsAllls.pop((chrx,pos))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
288 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
289 pos=endExs[-1]+100#max pos of exns
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
290 elif pos>exEnd-1:# or len(dChrPosdPopsAllls)==0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
291 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
292 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
293 dPopsFasta[echPop]+=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
294 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
295 dPopsFasta[echPop]=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
296 cntExns+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
297 tmpAcmltdPos+=len(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
298 if len(dexnIntrvlSeq)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
299 exStart,exEnd=sorted(dexnIntrvlSeq.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
300 seq=dexnIntrvlSeq.pop((exStart,exEnd))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
301 tmpdPopsFasta=dict([(x,list(seq)) for x in dPopsAllls])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
302 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
303 break
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
304 if ldexnIntrvlSeq!=cntExns:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
305 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
306 dPopsFasta[echPop]+=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
307 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
308 lchrStartexEndpos=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
309 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
310 dPopsFasta=dict([(echPop,revseq(dPopsFasta[echPop])) for echPop in dPopsFasta])#[echPop]+=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
311 for ePos in strePos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
312 lchrStartexEndpos.append('\t'.join([ENSEMBLT,chrx,str(tmpAcmltdPos-ePos-1),str(dStrePosAbsPos[ePos])]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
313 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
314 for ePos in strePos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
315 lchrStartexEndpos.append('\t'.join([ENSEMBLT,chrx,str(ePos),str(dStrePosAbsPos[ePos])]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
316 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
317 return dPopsFasta,lchrStartexEndpos
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
318
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
319 def rtrnSeqVars(dENSEMBLTseqChrStEnEx,dENSEMBLTChrPosdAlls):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
320 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
321 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
322 dENSEMBLTPopsFasta={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
323 lchrStartexEndposAll=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
324 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
325 sENSEMBLTcmmn=set(dENSEMBLTChrPosdAlls.keys()).intersection(set(dENSEMBLTseqChrStEnEx.keys()))#sENSEMBLTcmmn between UCSC and ENSEMBLE
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
326 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
327 for ENSEMBLT in sENSEMBLTcmmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
328 seq,chrx,startExs,endExs,rvrse=dENSEMBLTseqChrStEnEx[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
329 dChrPosdPopsAllls=dENSEMBLTChrPosdAlls[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
330 if len(startExs)>0 and len(endExs)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
331 dPopsFasta,lchrStartexEndpos=rtrndPopsFasta(seq,chrx,startExs,endExs,rvrse,dChrPosdPopsAllls,ENSEMBLT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
332 lchrStartexEndposAll.extend(lchrStartexEndpos)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
333 if dPopsFasta:#to correct a bug of the input table, in cases in which endExons<startExn (!). See ENSCAFT00000000145 (MC4R) in canFam2 for example.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
334 dENSEMBLTPopsFasta[ENSEMBLT]=dPopsFasta
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
335 return dENSEMBLTPopsFasta,lchrStartexEndposAll
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
336
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
337
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
338
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
339 def rtrnPhy(dPopsFasta,ENSEMBLT):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
340 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
341 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
342 dPopsFormPhy={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
343 for eachPop in dPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
344 hader='%s'%eachPop
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
345 #~ hader='>%s'%eachPop
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
346 seq=dPopsFasta[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
347 formtd='\t'.join([hader,seq])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
348 #~ formtd='\n'.join([hader,seq])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
349 dPopsFormPhy[eachPop]=formtd
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
350 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
351 return dPopsFormPhy,len(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
352
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
353 def wrapSeqsFasta(dENSEMBLTPopsFasta,sPopsIntrst):
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
354 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
355 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
356 ENSEMBLTKaKs=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
357 nonHeader=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
358 cnt=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
359 lENSEMBLT=len(dENSEMBLTPopsFasta)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
360 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
361 for ENSEMBLT in sorted(dENSEMBLTPopsFasta.keys()):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
362 cnt+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
363 dPopsFasta=dENSEMBLTPopsFasta[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
364 dPopsFormPhy,lenseq=rtrnPhy(dPopsFasta,ENSEMBLT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
365 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
366 seqPMLformat=['%s %s'%(len(dPopsFormPhy),lenseq)]#generate new PHYML sequence
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
367 #~ seqPMLformat=[]#generate new PHYML sequence
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
368 for namex in sorted(sPopsIntrst):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
369 seqPMLformat.append(dPopsFormPhy[namex])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
370 #~
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
371 outFastaf=open('%s.phy'%ENSEMBLT,'w')
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
372 outFastaf.write('\n'.join(seqPMLformat))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
373 outFastaf.close()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
374 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
375 return 0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
376
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
377 def pos_dict(gd_indivs_file, input_type):
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
378 rv = {}
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
379
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
380 p = Population()
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
381 p.from_population_file(gd_indivs_file)
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
382
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
383 for tag in p.tag_list():
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
384 column, name = tag.split(':')
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
385 column = int(column) - 1
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
386
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
387 if input_type == 'gd_genotype':
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
388 column -= 2
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
389
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
390 rv[name] = column
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
391
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
392 return rv
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
393
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
394 def main():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
395 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
396 #~bpython mkPhyl.py --input=colugo_mt_Galaxy_genotypes.txt --chrClmn=0 --posClmn=1 --refClmn=2 --altrClmn=3 --output=out.d --gd_indivs=genotypes.gd_indivs --inputCover=colugo_mt_Galaxy_coverage.txt --gd_indivs_cover=coverage.gd_indivs --cvrgTreshold=0 --chrClmnCvrg=0 --posClmnCvrg=1 --refClmnCvrg=2 --altrClmnCvrg=3 --indvlsPrctTrshld=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
397 parser = argparse.ArgumentParser(description='Returns the count of genes in KEGG categories and their statistical overrrepresentation, from a list of genes and an background file (i.e. plane text with ENSEMBLT and KEGG pathways).')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
398 parser.add_argument('--input',metavar='input gd_snp file',type=str,help='the input file with the table in gd_snp/gd_genotype format.',required=True)
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
399 parser.add_argument('--input_type',metavar='input type',type=str,help='the input file type (gd_snp or gd_genotype)',required=True)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
400 parser.add_argument('--chrClmn',metavar='int',type=int,help='the column with the chromosome.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
401 parser.add_argument('--posClmn',metavar='int',type=int,help='the column with the SNPs position.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
402 parser.add_argument('--refClmn',metavar='int',type=int,help='the column with the reference nucleotide.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
403 parser.add_argument('--altrClmn',metavar='int',type=int,help='the column with the derived nucleotide.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
404 parser.add_argument('--output',metavar='output',type=str,help='the output',required=True)
37
884ccb07885b Fixed output_id
Richard Burhans <burhans@bx.psu.edu>
parents: 36
diff changeset
405 parser.add_argument('--output_id',metavar='int',type=int,help='the output id',required=True)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
406 parser.add_argument('--gd_indivs',metavar='input gd_indivs file',type=str,help='the input reference species columns in the input file.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
407 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
408 parser.add_argument('--inputCover',metavar='input gd_snp cover file',type=str,help='the input file with the table in gd_snp/gd_genotype cover format.',required=False,default=False)
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
409 parser.add_argument('--inputCover_type',metavar='input cover type',type=str,help='the cover input file type (gd_snp or gd_genotype)',required=False,default=False)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
410 parser.add_argument('--gd_indivs_cover',metavar='input gd_indivs file',type=str,help='the input reference species columns in the input cover file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
411 parser.add_argument('--cvrgTreshold',metavar='input coverage threshold',type=int,help='the coverage threshold above which nucleotides are included, else "N".',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
412 parser.add_argument('--chrClmnCvrg',metavar='int',type=int,help='the column with the chromosome in the input coverage file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
413 parser.add_argument('--posClmnCvrg',metavar='int',type=int,help='the column with the SNPs position in the input coverage file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
414 parser.add_argument('--refClmnCvrg',metavar='int',type=int,help='the column with the reference nucleotide in the input coverage file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
415 parser.add_argument('--altrClmnCvrg',metavar='int',type=int,help='the column with the derived nucleotide in the input coverage file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
416 parser.add_argument('--indvlsPrctTrshld',metavar='int',type=float,help='the percentage of individual above which nucleotides are included, else "N".',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
417 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
418 parser.add_argument('--sequence',metavar='input fasta file',type=str,help='the input file with the sequence whose SNPs are in the input.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
419 parser.add_argument('--gene_info',metavar='input interval file',type=str,help='the input interval file with the the information on the genes.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
420 parser.add_argument('--fchrClmn',metavar='int',type=int,help='the column with the chromosome in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
421 parser.add_argument('--txStartClmn',metavar='int',type=int,help='the column with the transcript start column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
422 parser.add_argument('--txEndClmn',metavar='int',type=int,help='the column with the transcript end column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
423 parser.add_argument('--strandClmn',metavar='int',type=int,help='the column with the strand column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
424 parser.add_argument('--geneNameClmn',metavar='int',type=int,help='the column with the gene name column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
425 parser.add_argument('--cdsStartClmn',metavar='int',type=int,help='the column with the coding start column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
426 parser.add_argument('--cdsEndClmn',metavar='int',type=int,help='the column with the coding end column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
427 parser.add_argument('--startExsClmn',metavar='int',type=int,help='the column with the exon start positions column in the gene_info file.',required=False,default=False)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
428 parser.add_argument('--endExsClmn',metavar='int',type=int,help='the column with the exon end positions column in the gene_info file.',required=False,default=False)
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
429
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
430 args = parser.parse_args()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
431
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
432 inSNPf = args.input
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
433 inSNPf_type = args.input_type
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
434 outfile = args.output
37
884ccb07885b Fixed output_id
Richard Burhans <burhans@bx.psu.edu>
parents: 36
diff changeset
435 outfile_id = args.output_id
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
436 gd_indivs = args.gd_indivs
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
437 pxchrx = args.chrClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
438 pxpos = args.posClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
439 pxntA = args.refClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
440 pxntB = args.altrClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
441
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
442
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
443 inCDSfile = args.sequence
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
444 inUCSCfile = args.gene_info
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
445 fchrClmn = args.fchrClmn#chromosome column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
446 txStartClmn = args.txStartClmn#transcript start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
447 txEndClmn = args.txEndClmn#transcript end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
448 strandClmn = args.strandClmn#strand column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
449 geneNameClmn = args.geneNameClmn#gene name column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
450 cdsStartClmn = args.cdsStartClmn#coding sequence start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
451 cdsEndClmn = args.cdsEndClmn#coding sequence end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
452 startExsClmn = args.startExsClmn#exons start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
453 endExsClmn = args.endExsClmn#exons end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
454
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
455 inputCover = args.inputCover
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
456 inputCover_type = args.inputCover_type
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
457 gd_indivs_cover = args.gd_indivs_cover
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
458 cvrgTreshold = args.cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
459 pxchrxCov = args.chrClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
460 pxposCov = args.posClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
461 pxntACov = args.refClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
462 pxntBCov = args.altrClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
463 indvlsPrctTrshld = args.indvlsPrctTrshld
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
464
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
465
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
466 #print inputCover, gd_indivs_cover, cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
467
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
468 assert ((inputCover and gd_indivs_cover and cvrgTreshold>=0 and indvlsPrctTrshld>=0) or (inCDSfile and inUCSCfile))
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
469
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
470 dPopsinSNPfPos = pos_dict(gd_indivs, inSNPf_type)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
471 #~ dPopsinSNPfPos.update({'ref':False})
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
472 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
473 sPopsIntrst=set(dPopsinSNPfPos.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
474 dChrdPosdPopsAlllsInit,seqref,chrx,startExs,endExs=rtrnFxdChrPos(inSNPf,dPopsinSNPfPos,pxchrx,pxpos,pxntA,pxntB)#~ print '1. Getting fixed alleles information...'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
475 #~ dENSEMBLTseqChrStEnEx,dChrdStrtEndENSEMBLT=rtrndENSEMBLTseq(inCDSfile,inUCSCfile)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
476 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
477 if inputCover and gd_indivs_cover and cvrgTreshold>=0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
478 dPopsinSNPfPos_cover=dict([(eachPop,False) for eachPop in dPopsinSNPfPos.keys()])
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
479 dPopsinSNPfPos_cover.update(pos_dict(gd_indivs_cover, inputCover_type))
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
480 dChrdPosdPopsAlllsInit,seqref,chrx,startExs,endExs=rtrnFxdChrPos(inputCover,dPopsinSNPfPos_cover,pxchrxCov,pxposCov,pxntACov,pxntBCov,dChrdPosdPopsAlllsInit,cvrgTreshold,indvlsPrctTrshld)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
481 rvrse=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
482 dENSEMBLTseqChrStEnEx={'tmp':(seqref,chrx,startExs,endExs,rvrse)}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
483 dChrdStrtEndENSEMBLT={chrx:{(startExs[0],endExs[0]):'tmp'}}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
484 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
485 elif inCDSfile and inUCSCfile:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
486 dENSEMBLTseqChrStEnEx,dChrdStrtEndENSEMBLT=rtrndENSEMBLTseq(inCDSfile,inUCSCfile,fchrClmn,txStartClmn,txEndClmn,strandClmn,geneNameClmn,startExsClmn,endExsClmn,cdsStartClmn,cdsEndClmn)#~ print '2. Getting transcripts and exons information...'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
487 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
488 dENSEMBLTChrPosdAlls,dChrPosdPopsAllls=rtrnFxdChrPosinCodReg(dChrdStrtEndENSEMBLT,dChrdPosdPopsAlllsInit)#~ print '3. Getting fixed alleles in exons...'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
489 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
490 dENSEMBLTPopsFasta,lchrStartexEndposAll=rtrnSeqVars(dENSEMBLTseqChrStEnEx,dENSEMBLTChrPosdAlls)#~ print '4. Getting fasta sequences of populations...'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
491 #~
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
492 wrapSeqsFasta(dENSEMBLTPopsFasta,sPopsIntrst)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
493 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
494
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
495 ## get a list of output files
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
496 files = glob.glob('*.phy')
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
497
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
498 if len(files) == 0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
499 with open(outfile, 'w') as ofh:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
500 print >> ofh, 'No output.'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
501 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
502 ## the first file becomes the output
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
503 file = files.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
504 shutil.move(file, outfile)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
505
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
506 ## rename the rest of the files
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
507 for file in files:
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
508 name = file[:-4]
35
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
509 name = name.replace('_', '-')
ea52b23f1141 Bug fixes for Draw variants, Phylip, and gd_d_tools
Richard Burhans <burhans@bx.psu.edu>
parents: 31
diff changeset
510 new_filename = 'primary_{0}_{1}_visible_txt_?'.format(outfile_id, name)
36
51cd0307fb70 Phylip's extra ouputs are now stored in the job working directory
Richard Burhans <burhans@bx.psu.edu>
parents: 35
diff changeset
511 os.rename(file, new_filename)
31
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
512
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
513 return 0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
514
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
515 if __name__ == '__main__':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
516 main()