|
0
|
1 '''
|
|
|
2 get binned score of intervals,allow extension
|
|
|
3 '''
|
|
|
4
|
|
|
5 import os,sys,numpy,random,string
|
|
|
6
|
|
|
7 from resize import *
|
|
|
8
|
|
|
9 from bx.bbi.bigwig_file import BigWigFile
|
|
|
10
|
|
|
11 def binning(x,n):
|
|
|
12 # make n bin of x
|
|
|
13 y = numpy.zeros(n,dtype=float)
|
|
|
14 if len(x) == 0:
|
|
|
15 return y
|
|
|
16 step = float(len(x))/n
|
|
|
17 for k in range(n):
|
|
|
18 i = int(step*k)
|
|
|
19 j = int(step*(k+1)) + 1
|
|
|
20 y[k] = x[i:j].mean()
|
|
|
21 #print i,j,k,y[k]
|
|
|
22 return y
|
|
|
23
|
|
|
24 def getBinnedScore(bwfile,intvfile,outfile,outplot,nbin):
|
|
|
25 '''
|
|
|
26 get binned average and std
|
|
|
27 '''
|
|
|
28 fbw = open(bwfile)
|
|
|
29 bw = BigWigFile(file=fbw)
|
|
|
30 fin = open(intvfile)
|
|
|
31 out = open(outfile,'w')
|
|
|
32 zeros = '\t'.join(['0']*nbin)
|
|
|
33 for line in fin:
|
|
|
34 #chrom,start,end,name,score,strand
|
|
|
35 line = line.strip()
|
|
|
36 flds = line.split('\t')
|
|
|
37 #get the score at base resolution as an array
|
|
|
38 scores = bw.get_as_array(flds[0],int(flds[1]),int(flds[2]))
|
|
|
39 if scores == None:
|
|
|
40 print 'not found:\t',line
|
|
|
41 out.write(line+'\t'+zeros+'\n')
|
|
|
42 continue
|
|
|
43 # reverse if on minus strand
|
|
|
44 if flds[5] == '-':
|
|
|
45 scores = scores[::-1]
|
|
|
46 # no score = 0
|
|
|
47 scores = numpy.nan_to_num(scores)
|
|
|
48 # bin the data
|
|
|
49 binned = binning(scores,nbin)
|
|
|
50 out.write(line+'\t'+'\t'.join(map(str,binned))+'\n')
|
|
|
51 fin.close()
|
|
|
52 out.close()
|
|
|
53 # plot
|
|
|
54 if nbin > 1:
|
|
|
55 tmp = "".join(random.sample(string.letters+string.digits, 8))
|
|
|
56 rscript = open(tmp,"w")
|
|
|
57 rscript.write("options(warn=-1)\n")
|
|
|
58 rscript.write("x <- read.table('"+outfile+"',sep='\t')\n")
|
|
|
59 rscript.write("x <- x[,(ncol(x)+1-"+str(nbin)+"):ncol(x)]\n")
|
|
|
60 rscript.write("pdf('"+outplot+"')\n")
|
|
|
61 rscript.write("avg <- apply(x,2,mean)\n")
|
|
|
62 rscript.write("err <- apply(x,2,sd)/sqrt(nrow(x))\n")
|
|
|
63 rscript.write("print(avg)\n")
|
|
|
64 rscript.write("print(err)\n")
|
|
|
65 rscript.write("ylim=c(min(avg-err),max(avg+err))\n")
|
|
|
66 rscript.write("xticks <- seq(ncol(x))\n")
|
|
|
67 rscript.write("plot(xticks,avg,xlab='',ylab='average',type='l',lwd=0,ylim=ylim)\n")
|
|
|
68 rscript.write("polygon(c(xticks,rev(xticks)),c(avg+err,rev(avg-err)),col='lightgreen',border=NA)\n")
|
|
|
69 rscript.write("lines(xticks,avg,type='l',lwd=1)\n")
|
|
|
70 rscript.write("dev.off()\n")
|
|
|
71 rscript.close()
|
|
|
72 os.system("R --vanilla < "+tmp)
|
|
|
73 os.system("rm "+tmp)
|
|
|
74
|
|
|
75 print sys.argv
|
|
|
76 prog,bwfile,intvfile,nbin,outfile,outplot = sys.argv
|
|
|
77 getBinnedScore(bwfile,intvfile,outfile,outplot,int(nbin))
|
