annotate cluster_onConnctdComps.py @ 32:03c22b722882

remove BeautifulSoup dependency
author Richard Burhans <burhans@bx.psu.edu>
date Fri, 20 Sep 2013 13:54:23 -0400
parents a631c2f6d913
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 # Cluster_GOKEGG.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 os
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
25 from networkx import connected_components,Graph,clustering
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
26 from numpy import percentile
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
27 from decimal import Decimal,getcontext
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
28 from itertools import permutations,combinations
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
29 import sys
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
30
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
31 def rtrnClustrsOnCltrCoff(dNodesWeightMin,threshold,perctile=True):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
32 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
33 From a file with three columns: nodeA, nodeB and a score, it returns
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
34 the strong and weak connected components produced when the edges
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
35 below the percentage threshold (or value) are excluded.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
36 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
37 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
38 Gmin = Graph()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
39 for nodeA,nodeB in dNodesWeightMin:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
40 wMin=dNodesWeightMin[nodeA,nodeB]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
41 Gmin.add_edge(nodeA,nodeB,weight=wMin)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
42 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
43 clstrCoffcMin=clustering(Gmin,weight='weight')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
44 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
45 if perctile:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
46 umbralMin=percentile(clstrCoffcMin.values(),threshold)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
47 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
48 umbralMin=threshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
49 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
50 GminNdsRmv=[x for x in clstrCoffcMin if clstrCoffcMin[x]<umbralMin]
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 Gmin.remove_nodes_from(GminNdsRmv)
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 dTermCmptNumbWkMin=rtrndata(Gmin)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
55 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
56 salelClustr=[]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
57 srtdterms=sorted(dTermCmptNumbWkMin.keys())
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
58 for echTerm in srtdterms:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
59 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
60 MinT=dTermCmptNumbWkMin[echTerm]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
61 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
62 MinT='-'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
63 salelClustr.append('\t'.join([echTerm,MinT]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
64 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
65 return salelClustr
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
66
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
67 def rtrndata(G):
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 returna list of terms and its clustering, as well as clusters from
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
70 a networkx formatted file.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
71 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
72 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
73 cntCompnts=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
74 dTermCmptNumbWk={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
75 for echCompnt in connected_components(G):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
76 cntCompnts+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
77 #print '.'.join(echCompnt)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
78 for echTerm in echCompnt:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
79 dTermCmptNumbWk[echTerm]=str(cntCompnts)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
80 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
81 return dTermCmptNumbWk
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
82
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
83 def rtrnCATcENSEMBLc(inCATfile,classClmns,ENSEMBLTcolmn,nonHdr=True):
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 return a dictionary of all the categories in an input file with
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
86 a set of genes. Takes as input a file with categories an genes.
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 dCAT={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
89 dENSEMBLTCAT={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
90 for eachl in open(inCATfile,'r'):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
91 if nonHdr and eachl.strip():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
92 ENSEMBLT=eachl.splitlines()[0].split('\t')[ENSEMBLTcolmn]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
93 sCAT=set()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
94 for CATcolmn in classClmns:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
95 sCAT.update(set([x for x in eachl.splitlines()[0].split('\t')[CATcolmn].split('.')]))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
96 sCAT=sCAT.difference(set(['','U','N']))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
97 if len(sCAT)>0:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
98 dENSEMBLTCAT[ENSEMBLT]=sCAT
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
99 for CAT in sCAT:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
100 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
101 dCAT[CAT].add(ENSEMBLT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
102 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
103 dCAT[CAT]=set([ENSEMBLT])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
104 nonHdr=True
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
105 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
106 dCAT=dict([(x,len(dCAT[x])) for x in dCAT.keys()])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
107 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
108 return dCAT,dENSEMBLTCAT
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
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
111 def calcDistance(sCAT1,sCAT2):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
112 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
113 takes as input two set of genesin different categories and returns
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
114 a value 1-percentage of gene shared cat1->cat2, and cat2->cat1.
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
115 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
116 getcontext().prec=5
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
117 lgensS1=Decimal(len(sCAT1))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
118 lgensS2=Decimal(len(sCAT2))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
119 shrdGns=sCAT1.intersection(sCAT2)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
120 lenshrdGns=len(shrdGns)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
121 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
122 dC1C2=1-(lenshrdGns/lgensS1)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
123 dC2C1=1-(lenshrdGns/lgensS2)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
124 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
125 return dC1C2,dC2C1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
126
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
127 def rtnPrwsdtncs(dCAT,dENSEMBLTCAT):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
128 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
129 return a mcl formated pairwise distances from a list of categories
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
130 """
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
131 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
132 getcontext().prec=5
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
133 dCATdst={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
134 lENSEMBL=dENSEMBLTCAT.keys()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
135 l=len(lENSEMBL)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
136 c=0
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
137 for ENSEMBL in lENSEMBL:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
138 c+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
139 lCAT=dENSEMBLTCAT.pop(ENSEMBL)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
140 for CAT1,CAT2 in combinations(lCAT, 2):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
141 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
142 dCATdst[CAT1,CAT2]+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
143 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
144 dCATdst[CAT1,CAT2]=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
145 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
146 dCATdst[CAT2,CAT1]+=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
147 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
148 dCATdst[CAT2,CAT1]=1
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
149 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
150 dNodesWeightMin={}
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
151 l=len(dCATdst)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
152 for CAT1,CAT2 in dCATdst.keys():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
153 shrdGns=dCATdst.pop((CAT1,CAT2))
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
154 dC1C2=float(shrdGns)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
155 nodeA,nodeB=sorted([CAT1,CAT2])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
156 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
157 cscor=dNodesWeightMin[nodeA,nodeB]
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
158 if cscor>=dC1C2:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
159 dNodesWeightMin[nodeA,nodeB]=dC1C2
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
160 except:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
161 dNodesWeightMin[nodeA,nodeB]=dC1C2
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
162 #
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
163 return dNodesWeightMin
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
164
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
165 def parse_class_columns(val, max_col):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
166 int_list = []
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 for elem in [x.strip() for x in val.split(',')]:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
169 if elem[0].lower() != 'c':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
170 print >> sys.stderr, "bad column format:", elem
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
171 sys.exit(1)
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 int_val = as_int(elem[1:])
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
174
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
175 if int_val is None:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
176 print >> sys.stderr, "bad column format:", elem
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
177 sys.exit(1)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
178 elif not 1 <= int_val <= max_col:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
179 print >> sys.stderr, "column out of range:", elem
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
180 sys.exit(1)
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 int_list.append(int_val - 1)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
183
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
184 return int_list
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
185
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
186 def as_int(val):
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
187 try:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
188 return int(val)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
189 except ValueError:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
190 return None
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
191 else:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
192 raise
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
193
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
194 def main():
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
195 """
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 #~ bpython cluster_onConnctdComps.py --input=../conctFinal_CEU.tsv --outfile=../borrar.txt --threshold=90 --ENSEMBLTcolmn=1 --classClmns='20 22'
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
198 parser = argparse.ArgumentParser(description='Returns the count of genes in ...')
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
199 parser.add_argument('--input',metavar='input TXT file',type=str,help='the input file with the table in txt format.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
200 parser.add_argument('--input_columns',metavar='input INT value',type=int,help='the number of columns in the input file.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
201 parser.add_argument('--outfile',metavar='input TXT file',type=str,help='the output file with the connected components.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
202 parser.add_argument('--threshold',metavar='input FLOAT value',type=float,help='the threshold to disconnect the nodes.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
203 parser.add_argument('--ENSEMBLTcolmn',metavar='input INT file',type=int,help='the column with the ENSEMBLE code in the input.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
204 parser.add_argument('--classClmns',metavar='input STR value',type=str,help='the list of columns with the gene categories separated by space.',required=True)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
205 args = parser.parse_args()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
206 infile = args.input
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
207 threshold = args.threshold
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
208 outfile = args.outfile
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
209 ENSEMBLTcolmn = args.ENSEMBLTcolmn
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
210 classClmns = parse_class_columns(args.classClmns, args.input_columns)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
211 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
212 dCAT,dENSEMBLTCAT=rtrnCATcENSEMBLc(infile,classClmns,ENSEMBLTcolmn)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
213 dNodesWeightMin=rtnPrwsdtncs(dCAT,dENSEMBLTCAT)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
214 salelClustr=rtrnClustrsOnCltrCoff(dNodesWeightMin,threshold)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
215 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
216 with open(outfile, 'w') as salef:
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
217 print >> salef, '\n'.join(salelClustr)
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
218 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
219 #~
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
220
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
221 if __name__ == '__main__':
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
222 main()
a631c2f6d913 Update to Miller Lab devshed revision 3c4110ffacc3
Richard Burhans <burhans@bx.psu.edu>
parents:
diff changeset
223