|
11
|
1 '''
|
|
|
2 get binned score of intervals,allow extension
|
|
|
3 '''
|
|
|
4
|
|
|
5 import os,sys,numpy
|
|
|
6 import string, random
|
|
|
7
|
|
|
8 from resize import *
|
|
|
9
|
|
|
10 from bx.bbi.bigwig_file import BigWigFile
|
|
|
11
|
|
|
12 def binning(x,n):
|
|
|
13 # make n bin of x
|
|
|
14 y = numpy.zeros(n,dtype=float)
|
|
|
15 if len(x) == 0:
|
|
|
16 return y
|
|
|
17 step = float(len(x))/n
|
|
|
18 for k in range(n):
|
|
|
19 i = int(step*k)
|
|
|
20 j = int(step*(k+1)) + 1
|
|
|
21 y[k] = x[i:j].mean()
|
|
|
22 #print i,j,k,y[k]
|
|
|
23 return y
|
|
|
24
|
|
|
25 def getBinnedScore(bwfile,intvfile,nbin):
|
|
|
26 '''
|
|
|
27 get binned average and std
|
|
|
28 '''
|
|
|
29 fbw = open(bwfile)
|
|
|
30 bw = BigWigFile(file=fbw)
|
|
|
31 fin = open(intvfile)
|
|
|
32 avg = numpy.zeros(nbin)
|
|
|
33 sqr = numpy.zeros(nbin)
|
|
|
34 N = 0
|
|
|
35 for line in fin:
|
|
|
36 #print N
|
|
|
37 #chrom,start,end,name,score,strand
|
|
|
38 flds = line.strip().split('\t')
|
|
|
39 #get the score at base resolution as an array
|
|
|
40 scores = bw.get_as_array(flds[0],int(flds[1]),int(flds[2]))
|
|
|
41 if scores == None:
|
|
|
42 print 'not found:\t',N,line
|
|
|
43 continue
|
|
|
44 N = N + 1
|
|
|
45 #print line,scores
|
|
|
46 # reverse if on minus strand
|
|
|
47 if flds[5] == '-':
|
|
|
48 scores = scores[::-1]
|
|
|
49 # no score = 0
|
|
|
50 scores = numpy.nan_to_num(scores)
|
|
|
51 # bin the data
|
|
|
52 binned = binning(scores,nbin)
|
|
|
53 avg = avg + binned
|
|
|
54 sqr = sqr + binned**2
|
|
|
55 # compute avg and std
|
|
|
56 avg = avg / N
|
|
|
57 err = ((sqr/N-avg**2)**0.5)/(N**0.5)
|
|
|
58 return avg,err,N
|
|
|
59
|
|
|
60 def getExtendedBinScore(bwfile,intvfile,nbins,exts):
|
|
|
61 '''
|
|
|
62 nbins: n1,n2,n3
|
|
|
63 exts: l1,l2,l3,l4
|
|
|
64 '''
|
|
|
65 avg = []
|
|
|
66 err = []
|
|
|
67 tmpfile = "".join(random.sample(string.letters+string.digits, 8))
|
|
|
68 if exts[0]>0 or exts[1]>0:
|
|
|
69 print 'make left extension'
|
|
|
70 resize(intvfile,tmpfile,'start-'+str(exts[0]),'start+'+str(exts[1]),'stranded')
|
|
|
71 print 'compute binned average'
|
|
|
72 avg,err,N = getBinnedScore(bwfile,tmpfile,nbins[0])
|
|
|
73 print 'regions used:',N
|
|
|
74 print 'make center region'
|
|
|
75 resize(intvfile,tmpfile,'start+'+str(exts[1]),'end-'+str(exts[2]),'stranded')
|
|
|
76 print 'compute binned average'
|
|
|
77 avg1,err1,N = getBinnedScore(bwfile,tmpfile,nbins[1])
|
|
|
78 print 'regions used:',N
|
|
|
79 avg = numpy.concatenate((avg,avg1))
|
|
|
80 err = numpy.concatenate((err,err1))
|
|
|
81 if exts[2]>0 or exts[3]>0:
|
|
|
82 print 'make right region'
|
|
|
83 resize(intvfile,tmpfile,'end-'+str(exts[2]),'end+'+str(exts[3]),'stranded')
|
|
|
84 print 'compute binned average'
|
|
|
85 avg2,err2,N = getBinnedScore(bwfile,tmpfile,nbins[2])
|
|
|
86 print 'regions used:',N
|
|
|
87 avg = numpy.concatenate((avg,avg2))
|
|
|
88 err = numpy.concatenate((err,err2))
|
|
|
89 os.system('rm '+tmpfile)
|
|
|
90 return avg,err
|
|
|
91
|
|
|
92 prog,bwfile,intvfile,exts,nbins,outfile,outplot = sys.argv
|
|
|
93 nbins = numpy.fromstring(nbins,dtype=int,sep=',')
|
|
|
94 exts = numpy.fromstring(exts,dtype=int,sep=',')
|
|
|
95 avg, err = getExtendedBinScore(bwfile,intvfile,nbins,exts)
|
|
|
96 print 'save data'
|
|
|
97 out = open(outfile,'w')
|
|
|
98 numpy.savetxt(out, avg, fmt='%.6f', delimiter=' ', newline=' ')
|
|
|
99 out.write('\n')
|
|
|
100 numpy.savetxt(out, err, fmt='%.6f', delimiter=' ', newline=' ')
|
|
|
101 out.write('\n')
|
|
|
102 out.close()
|
|
|
103
|
|
|
104 print 'plot'
|
|
|
105 start = exts[0]*nbins[0]/(exts[0]+exts[1])
|
|
|
106 end = nbins[0]+nbins[1]+exts[2]*nbins[2]/(exts[2]+exts[3])
|
|
|
107 #print start,end
|
|
|
108 rscript = open("tmp.r","w")
|
|
|
109 rscript.write("options(warn=-1)\n")
|
|
|
110 rscript.write("x <- read.table('"+outfile+"')\n")
|
|
|
111 rscript.write("pdf('"+outplot+"')\n")
|
|
|
112 rscript.write("avg <- x[1,]\n")
|
|
|
113 rscript.write("err <- x[2,]\n")
|
|
|
114 #rscript.write("print(x)\n")
|
|
|
115 rscript.write("ylim=c(min(avg-err),max(avg+err))\n")
|
|
|
116 rscript.write("xticks <- seq(ncol(x))\n")
|
|
|
117 #rscript.write("print(xticks)\n")
|
|
|
118 rscript.write("plot(xticks,avg,xlab='',ylab='average coverage',type='l',lwd=0,ylim=ylim,xaxt='n')\n")
|
|
|
119 rscript.write("axis(1, at=c(min(xticks),"+str(start)+","+str(end)+",max(xticks)),labels=c(-"+str(exts[0])+",0,0,"+str(exts[3])+"), las=2)\n")
|
|
|
120 rscript.write("polygon(c(xticks,rev(xticks)),c(avg+err,rev(avg-err)),col='lightgreen',border=NA)\n")
|
|
|
121 rscript.write("lines(xticks,avg,type='l',lwd=1)\n")
|
|
|
122 rscript.write("lines(c(min(xticks),max(xticks)),c(0,0),lwd=2)\n")
|
|
|
123 rscript.write("lines(c("+str(start)+","+str(end)+"),c(0,0),lwd=10)\n")
|
|
|
124 rscript.write("dev.off()\n")
|
|
|
125 rscript.close()
|
|
|
126 os.system("R --vanilla --slave < tmp.r")
|
|
|
127 os.system("rm tmp.r")
|
|
|
128
|
