comparison tools/mytools/binnedAverage.py @ 0:9071e359b9a3

Uploaded

0:9071e359b9a3

'''
get binned score of intervals,allow extension
'''

import os,sys,numpy,random,string

from resize import *

from bx.bbi.bigwig_file import BigWigFile

def binning(x,n):
    # make n bin of x
    y = numpy.zeros(n,dtype=float)
    if len(x) == 0:
        return y
    step = float(len(x))/n
    for k in range(n):
        i = int(step*k)
        j = int(step*(k+1)) + 1
        y[k] = x[i:j].mean()
        #print i,j,k,y[k]
    return y

def getBinnedScore(bwfile,intvfile,outfile,outplot,nbin):
    '''
    get binned average and std
    '''
    fbw = open(bwfile)
    bw = BigWigFile(file=fbw)
    fin = open(intvfile)
    out = open(outfile,'w')
    zeros = '\t'.join(['0']*nbin)
    for line in fin:
        #chrom,start,end,name,score,strand
        line = line.strip()
        flds = line.split('\t')
        #get the score at base resolution as an array
        scores = bw.get_as_array(flds[0],int(flds[1]),int(flds[2]))
        if scores == None:
            print 'not found:\t',line
            out.write(line+'\t'+zeros+'\n')
            continue
        # reverse if on minus strand
        if flds[5] == '-':
            scores = scores[::-1]
        # no score = 0    
        scores = numpy.nan_to_num(scores)
        # bin the data
        binned = binning(scores,nbin)
        out.write(line+'\t'+'\t'.join(map(str,binned))+'\n')
    fin.close()
    out.close()
    # plot
    if nbin > 1:
        tmp = "".join(random.sample(string.letters+string.digits, 8))
        rscript = open(tmp,"w")
        rscript.write("options(warn=-1)\n")
        rscript.write("x <- read.table('"+outfile+"',sep='\t')\n")
        rscript.write("x <- x[,(ncol(x)+1-"+str(nbin)+"):ncol(x)]\n")
        rscript.write("pdf('"+outplot+"')\n")
        rscript.write("avg <- apply(x,2,mean)\n")
        rscript.write("err <- apply(x,2,sd)/sqrt(nrow(x))\n")
        rscript.write("print(avg)\n")
        rscript.write("print(err)\n")
        rscript.write("ylim=c(min(avg-err),max(avg+err))\n")
        rscript.write("xticks <- seq(ncol(x))\n")
        rscript.write("plot(xticks,avg,xlab='',ylab='average',type='l',lwd=0,ylim=ylim)\n")   
        rscript.write("polygon(c(xticks,rev(xticks)),c(avg+err,rev(avg-err)),col='lightgreen',border=NA)\n")
        rscript.write("lines(xticks,avg,type='l',lwd=1)\n")   
        rscript.write("dev.off()\n")
        rscript.close()
        os.system("R --vanilla < "+tmp)
        os.system("rm "+tmp)

print sys.argv
prog,bwfile,intvfile,nbin,outfile,outplot = sys.argv
getBinnedScore(bwfile,intvfile,outfile,outplot,int(nbin))