annotate make_phylip.py @ 31:a631c2f6d913

Update to Miller Lab devshed revision 3c4110ffacc3
author Richard Burhans <burhans@bx.psu.edu>
date Fri, 20 Sep 2013 13:25:27 -0400
parents
children ea52b23f1141
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
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
24 import errno
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
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
27
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 def mkdir_p(path):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30 os.makedirs(path)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31 except OSError, e:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 if e.errno <> errno.EEXIST:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 raise
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 revseq(seq):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 seq=list(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 seq.reverse()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38 seq=''.join(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39 return seq
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41 def revComp(allPop):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 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
43 allPopsComp=dAllCompAll[allPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44 return allPopsComp
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 rtrnCons(ntA,ntB):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 srtdPairs=''.join(sorted([ntA,ntB]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 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
49 cons=dpairsCons[srtdPairs]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 return cons
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
51
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
52 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
53 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
54 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 dChrdPosdPopsAlllsInit={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56 seqref=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 for eachl in open(inSNPf,'r'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 if eachl.strip() and eachl[0]!='#':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 fllInfoSplt=eachl.splitlines()[0].split('\t')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60 chrx=fllInfoSplt[pxchrx]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 pos=int(fllInfoSplt[pxpos])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62 ntA=fllInfoSplt[pxntA]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63 ntB=fllInfoSplt[pxntB]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64 seqref.append([pos,ntA])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
65 dPopsAllls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
66 if fulldChrdPosdPopsAlllsInit:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
68 cntIndv=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
69 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
70 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
71 fulldPopsAllls=fulldChrdPosdPopsAlllsInit[chrx][pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
72 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
73 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
74 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
75 for eachPop in dPopsinSNPfPos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
76 clmnCvrg=dPopsinSNPfPos[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
77 if clmnCvrg:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
78 eachPopCvrg=int(fllInfoSplt[clmnCvrg])
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 #~ eachPopCvrg=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
81 eachPopCvrg=cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
82 if eachPopCvrg>=cvrgTreshold:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
83 dPopsAllls[eachPop]=fulldPopsAllls[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
84 cntIndv+=1
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 dPopsAllls[eachPop]='N'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
87 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
88 if indvlsPrctTrshld>(cntIndv/float(len(dPopsinSNPfPos))):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
89 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
90 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
91 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
92 for eachPop in dPopsinSNPfPos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
93 if dPopsinSNPfPos[eachPop]:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
94 eachPopAll=int(fllInfoSplt[dPopsinSNPfPos[eachPop]])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
95 if eachPopAll==0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
96 dPopsAllls[eachPop]=ntB
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
97 elif eachPopAll==2:
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 elif eachPopAll==1:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
100 dPopsAllls[eachPop]=rtrnCons(ntA,ntB)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
101 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
102 dPopsAllls[eachPop]='N'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
103 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
104 dPopsAllls[eachPop]=ntA
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
105 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
106 dChrdPosdPopsAlllsInit[chrx][pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
107 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
108 dChrdPosdPopsAlllsInit[chrx]={pos:dPopsAllls}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
109 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
110 seqref.sort()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
111 startExs=[seqref[0][0]]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
112 endExs=[seqref[-1][0]+1]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
113 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
114 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
115 return dChrdPosdPopsAlllsInit,seqref,chrx,startExs,endExs
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
116
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
117
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
118 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
119 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
120 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
121 dENSEMBLTchrxStEndEx={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
122 dChrdStrtEndENSEMBLT={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
123 for eachl in open(inUCSCfile,'r'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
124 if eachl.strip():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
125 rvrse=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
126 allVls=eachl.split('\t')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
127 txStart=allVls[txStartClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
128 txEnd=allVls[txEndClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
129 ENSEMBLT=allVls[geneNameClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
130 strand=allVls[strandClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
131 chrx=allVls[fchrClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
132 if cdsStartClmn and cdsEndClmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
133 cdsStart=allVls[cdsStartClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
134 cdsEnd=allVls[cdsEndClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
135 if startExsClmn and endExsClmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
136 startExs=allVls[startExsClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
137 endExs=allVls[endExsClmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
138 if strand=='-':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
139 rvrse=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
140 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
141 dChrdStrtEndENSEMBLT[chrx][int(txStart),int(txEnd)]=ENSEMBLT
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
142 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
143 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
144 dChrdStrtEndENSEMBLT[chrx]={(int(txStart),int(txEnd)):ENSEMBLT}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
145 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
146 dChrdStrtEndENSEMBLT={chrx:{(int(txStart),int(txEnd)):ENSEMBLT}}
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 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
149 startExs,endExs=rtrnExnStarEndCorrc(startExs,endExs,cdsStart,cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
150 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
151 startExs,endExs=[int(txStart)],[int(txEnd)]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
152 dENSEMBLTchrxStEndEx[ENSEMBLT]=(chrx,startExs,endExs,rvrse)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
153 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
154 dENSEMBLTseq={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
155 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
156 for ENSEMBLT,seq in ENSEMBLTseqs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
157 dENSEMBLTseq[ENSEMBLT]=seq
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
158 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
159 dENSEMBLTseqChrStEnEx={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
160 for ENSEMBLT in dENSEMBLTchrxStEndEx:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
161 chrx,startExs,endExs,rvrse=dENSEMBLTchrxStEndEx[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
162 addEseqChrStEnEx=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
163 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
164 seq=dENSEMBLTseq[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
165 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
166 seq=revseq(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
167 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
168 addEseqChrStEnEx=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
169 if addEseqChrStEnEx:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
170 dENSEMBLTseqChrStEnEx[ENSEMBLT]=(seq,chrx,startExs,endExs,rvrse)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
171 return dENSEMBLTseqChrStEnEx,dChrdStrtEndENSEMBLT
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
172
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
173
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
174 def rtrnFxdChrPosinCodReg(dChrdStrtEndENSEMBLT,dChrdPosdPopsAlllsInit):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
175 """
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 dENSEMBLTChrPosdAlls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
178 dChrPosdPopsAllls={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
179 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
180 for x in todel:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
181 x=dChrdPosdPopsAlllsInit.pop(x)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
182 #---
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
183 while len(dChrdPosdPopsAlllsInit)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
184 chrx=dChrdPosdPopsAlllsInit.keys()[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
185 dStrtEndENSEMBLT=dChrdStrtEndENSEMBLT.pop(chrx)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
186 dPosdPopsAllls=dChrdPosdPopsAlllsInit.pop(chrx)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
187 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
188 srtdStrtEndENSEMBLT=sorted(dStrtEndENSEMBLT.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
189 srtdPosdPopsAllls=sorted(dPosdPopsAllls.keys())
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 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
192 strt,end=srtdStrtEndENSEMBLT.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
193 ENSEMBLT=dStrtEndENSEMBLT[strt,end]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
194 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
195 keePloop=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
196 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
197 while keePloop:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
198 if strt<=pos<=end:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
199 for tmpstrt,tmpend in [(strt,end)]+srtdStrtEndENSEMBLT:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
200 if tmpstrt<=pos<=tmpend:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
201 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
202 dChrPosdPopsAllls[chrx,pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
203 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
204 dENSEMBLTChrPosdAlls[ENSEMBLT][chrx,pos]=dPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
205 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
206 dENSEMBLTChrPosdAlls[ENSEMBLT]={(chrx,pos):dPopsAllls}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
207 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
208 continue
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
209 if len(srtdPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
210 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
211 dPopsAllls=dPosdPopsAllls[pos]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
212 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
213 keePloop=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
214 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
215 elif pos<=strt:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
216 if len(srtdPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
217 pos=srtdPosdPopsAllls.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
218 dPopsAllls=dPosdPopsAllls[pos]
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 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
222 if len(srtdStrtEndENSEMBLT)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
223 strt,end=srtdStrtEndENSEMBLT.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
224 ENSEMBLT=dStrtEndENSEMBLT[strt,end]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
225 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
226 keePloop=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
227 return dENSEMBLTChrPosdAlls,dChrPosdPopsAllls
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
228
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
229 def rtrnExnStarEndCorrc(startExs,endExs,cdsStart,cdsEnd):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
230 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
231 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
232 cdsStart,cdsEnd=int(cdsStart),int(cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
233 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
234 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
235 crrctdstartExs.add(cdsStart)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
236 crrctdendExs.add(cdsEnd)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
237 sStartDel=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
238 sEndDel=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
239 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
240 for echvl in crrctdstartExs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
241 if echvl<cdsStart or echvl>cdsEnd:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
242 sStartDel.add(echvl)
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 for echvl in crrctdendExs:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
245 if echvl<cdsStart or echvl>cdsEnd:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
246 sEndDel.add(echvl)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
247 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
248 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
249
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
250 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
251 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
252 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
253 exnIntrvl=zip(startExs,endExs)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
254 CDSinitPos=exnIntrvl[0][0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
255 dexnIntrvlSeq={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
256 for exStart,exEnd in exnIntrvl:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
257 lenEx=exEnd-exStart
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
258 dexnIntrvlSeq[exStart,exEnd]=seq[:lenEx]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
259 seq=seq[lenEx:]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
260
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
261 ldexnIntrvlSeq=len(dexnIntrvlSeq)
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 dPopsFasta={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
264 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
265 strePos=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
266 dStrePosAbsPos={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
267 tmpAcmltdPos=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
268 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
269 exStart,exEnd=sorted(dexnIntrvlSeq.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
270 seq=dexnIntrvlSeq.pop((exStart,exEnd))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
271 chrx,pos=sorted(dChrPosdPopsAllls.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
272 dPopsAllls=dChrPosdPopsAllls.pop((chrx,pos))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
273 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
274 cntExns=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
275 while True:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
276 if exStart<=pos<=exEnd-1:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
277 relPos=tmpAcmltdPos+pos-exStart
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
278 strePos.add(relPos)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
279 dStrePosAbsPos[relPos]=pos
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
280 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
281 allPop=dPopsAllls[echPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
282 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
283 allPop=revComp(allPop)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
284 tmpdPopsFasta[echPop][pos-exStart]=allPop
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<exStart:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
291 if len(dChrPosdPopsAllls)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
292 chrx,pos=sorted(dChrPosdPopsAllls.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
293 dPopsAllls=dChrPosdPopsAllls.pop((chrx,pos))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
294 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
295 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
296 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
297 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
298 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
299 dPopsFasta[echPop]+=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
300 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
301 dPopsFasta[echPop]=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
302 cntExns+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
303 tmpAcmltdPos+=len(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
304 if len(dexnIntrvlSeq)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
305 exStart,exEnd=sorted(dexnIntrvlSeq.keys())[0]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
306 seq=dexnIntrvlSeq.pop((exStart,exEnd))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
307 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
308 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
309 break
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
310 if ldexnIntrvlSeq!=cntExns:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
311 for echPop in tmpdPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
312 dPopsFasta[echPop]+=''.join(tmpdPopsFasta[echPop])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
313 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
314 lchrStartexEndpos=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
315 if rvrse:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
316 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
317 for ePos in strePos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
318 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
319 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
320 for ePos in strePos:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
321 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
322 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
323 return dPopsFasta,lchrStartexEndpos
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 def rtrnSeqVars(dENSEMBLTseqChrStEnEx,dENSEMBLTChrPosdAlls):
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 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
328 dENSEMBLTPopsFasta={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
329 lchrStartexEndposAll=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
330 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
331 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
332 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
333 for ENSEMBLT in sENSEMBLTcmmn:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
334 seq,chrx,startExs,endExs,rvrse=dENSEMBLTseqChrStEnEx[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
335 dChrPosdPopsAllls=dENSEMBLTChrPosdAlls[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
336 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
337 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
338 lchrStartexEndposAll.extend(lchrStartexEndpos)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
339 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
340 dENSEMBLTPopsFasta[ENSEMBLT]=dPopsFasta
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
341 return dENSEMBLTPopsFasta,lchrStartexEndposAll
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
342
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
343
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
344
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
345 def rtrnPhy(dPopsFasta,ENSEMBLT):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
346 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
347 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
348 dPopsFormPhy={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
349 for eachPop in dPopsFasta:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
350 hader='%s'%eachPop
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
351 #~ hader='>%s'%eachPop
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
352 seq=dPopsFasta[eachPop]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
353 formtd='\t'.join([hader,seq])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
354 #~ formtd='\n'.join([hader,seq])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
355 dPopsFormPhy[eachPop]=formtd
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
356 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
357 return dPopsFormPhy,len(seq)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
358
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
359 def wrapSeqsFasta(dENSEMBLTPopsFasta,outFastaFold,sPopsIntrst):
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 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
362 ENSEMBLTKaKs=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
363 nonHeader=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
364 cnt=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
365 lENSEMBLT=len(dENSEMBLTPopsFasta)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
366 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
367 for ENSEMBLT in sorted(dENSEMBLTPopsFasta.keys()):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
368 cnt+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
369 dPopsFasta=dENSEMBLTPopsFasta[ENSEMBLT]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
370 dPopsFormPhy,lenseq=rtrnPhy(dPopsFasta,ENSEMBLT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
371 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
372 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
373 #~ seqPMLformat=[]#generate new PHYML sequence
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
374 for namex in sorted(sPopsIntrst):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
375 seqPMLformat.append(dPopsFormPhy[namex])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
376 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
377 mkdir_p(outFastaFold)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
378 outFastaf=os.path.join(outFastaFold,'%s.phy'%ENSEMBLT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
379 outFastaf=open(outFastaf,'w')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
380 outFastaf.write('\n'.join(seqPMLformat))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
381 outFastaf.close()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
382 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
383 return 0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
384
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
385 def main():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
386 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
387 #~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
388 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
389 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)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
390 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
391 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
392 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
393 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
394 parser.add_argument('--output',metavar='output',type=str,help='the output',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
395 parser.add_argument('--output_id',metavar='int',type=int,help='the output id',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
396 parser.add_argument('--output_dir',metavar='output folder sequences',type=str,help='the output folder with the sequences.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
397 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
398 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
399 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)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
400 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
401 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
402 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
403 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
404 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
405 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
406 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
407 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
408 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
409 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
410 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
411 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
412 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
413 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
414 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
415 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
416 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
417 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
418 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)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
419
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
420 args = parser.parse_args()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
421
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
422 inSNPf = args.input
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
423 outfile = args.output
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
424 outfile_id = args.output_id
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
425 outFastaFold = './out'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
426 files_dir = args.output_dir
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
427 gd_indivs = args.gd_indivs
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
428 pxchrx = args.chrClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
429 pxpos = args.posClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
430 pxntA = args.refClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
431 pxntB = args.altrClmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
432
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
433
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
434 inCDSfile = args.sequence
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
435 inUCSCfile = args.gene_info
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
436 fchrClmn = args.fchrClmn#chromosome column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
437 txStartClmn = args.txStartClmn#transcript start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
438 txEndClmn = args.txEndClmn#transcript end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
439 strandClmn = args.strandClmn#strand column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
440 geneNameClmn = args.geneNameClmn#gene name column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
441 cdsStartClmn = args.cdsStartClmn#coding sequence start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
442 cdsEndClmn = args.cdsEndClmn#coding sequence end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
443 startExsClmn = args.startExsClmn#exons start column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
444 endExsClmn = args.endExsClmn#exons end column
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
445
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
446 inputCover = args.inputCover
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
447 gd_indivs_cover = args.gd_indivs_cover
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
448 cvrgTreshold = args.cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
449 pxchrxCov = args.chrClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
450 pxposCov = args.posClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
451 pxntACov = args.refClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
452 pxntBCov = args.altrClmnCvrg
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
453 indvlsPrctTrshld = args.indvlsPrctTrshld
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 #print inputCover, gd_indivs_cover, cvrgTreshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
456
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
457 assert ((inputCover and gd_indivs_cover and cvrgTreshold>=0 and indvlsPrctTrshld>=0) or (inCDSfile and inUCSCfile))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
458
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
459 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
460 dPopsinSNPfPos=dict([(x.split()[1],int(x.split()[0])-1) for x in open(gd_indivs).read().splitlines() if x.strip()])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
461 #~ dPopsinSNPfPos.update({'ref':False})
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
462 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
463 sPopsIntrst=set(dPopsinSNPfPos.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
464 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
465 #~ dENSEMBLTseqChrStEnEx,dChrdStrtEndENSEMBLT=rtrndENSEMBLTseq(inCDSfile,inUCSCfile)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
466 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
467 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
468 dPopsinSNPfPos_cover=dict([(eachPop,False) for eachPop in dPopsinSNPfPos.keys()])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
469 dPopsinSNPfPos_cover.update(dict([(x.split()[1],int(x.split()[0])-1) for x in open(gd_indivs_cover).read().splitlines() if x.strip()]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
470 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
471 rvrse=False
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
472 dENSEMBLTseqChrStEnEx={'tmp':(seqref,chrx,startExs,endExs,rvrse)}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
473 dChrdStrtEndENSEMBLT={chrx:{(startExs[0],endExs[0]):'tmp'}}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
474 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
475 elif inCDSfile and inUCSCfile:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
476 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
477 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
478 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
479 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
480 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
481 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
482 wrapSeqsFasta(dENSEMBLTPopsFasta,outFastaFold,sPopsIntrst)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
483 #~
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
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
486 ## get a lit of output files
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
487 files = []
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
488 for dirpath, dirnames, filenames in os.walk(outFastaFold):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
489 for file in filenames:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
490 if file.endswith('.phy'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
491 files.append( os.path.join(dirpath, file) )
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
492 del dirnames[:]
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 if len(files) == 0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
495 with open(outfile, 'w') as ofh:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
496 print >> ofh, 'No output.'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
497 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
498 ## the first file becomes the output
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
499 file = files.pop(0)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
500 shutil.move(file, outfile)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
501
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
502 ## rename/move the rest of the files
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
503 for i, file in enumerate(files):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
504 new_filename = 'primary_{0}_output{1}_visible_txt_?'.format(outfile_id, i+2)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
505 new_pathname = os.path.join(files_dir, new_filename)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
506 shutil.move(file, new_pathname)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
507
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
508 return 0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
509
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
510 if __name__ == '__main__':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
511 main()