annotate metaintv2.py @ 23:4e646baac551

Uploaded

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

import os,sys,numpy

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,nbin):
    '''
    get binned average and std
    '''
    fbw = open(bwfile)
    bw = BigWigFile(file=fbw)
    fin = open(intvfile)
    avg = numpy.zeros(nbin)
    sqr = numpy.zeros(nbin)
    N = 0
    for line in fin:
        #chrom,start,end,name,score,strand
        flds = line.strip().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
            continue
        N = N + 1
        #print line,scores
        # 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)
        avg = avg + binned
        sqr = sqr + binned**2
    # compute avg and std
    avg = avg / N
    err = ((sqr/N-avg**2)**0.5)/(N**0.5)
    return avg,err

def getExtendedBinScore(bwfile,intvfile,nbins,exts):
    '''
    nbins: n1,n2,n3
    exts: l1,l2,l3,l4
    '''
    # make left extension
    resize(intvfile,intvfile+'.tmp','start-'+str(exts[0]),'start+'+str(exts[1]),'stranded')
    # compute binned average
    avg,err = getBinnedScore(bwfile,intvfile+'.tmp',nbins[0])
    # make center region
    resize(intvfile,intvfile+'.tmp','start+'+str(exts[1]),'end-'+str(exts[2]),'stranded')
    # compute binned average
    avg1,err1 = getBinnedScore(bwfile,intvfile+'.tmp',nbins[1])    
    avg = numpy.concatenate((avg,avg1))
    err = numpy.concatenate((err,err1))
    # make right region
    resize(intvfile,intvfile+'.tmp','end-'+str(exts[2]),'end+'+str(exts[3]),'stranded')
    # compute binned average
    avg2,err2 = getBinnedScore(bwfile,intvfile+'.tmp',nbins[2])    
    avg = numpy.concatenate((avg,avg2))
    err = numpy.concatenate((err,err2))
    
    return avg,err

print sys.argv
bwfile,intvfile,exts,nbins,outfile,outplot = sys.argv
avg, err = getExtendedBinScore(bwfile,intvfile,numpy.fromstring(nbins,sep=','),numpy.fromstring(exts,sep=','))
out = open(outfile,'w')
numpy.savetxt(out, avg, fmt='%.6f', delimiter=' ', newline=' ')
out.write('\n')
numpy.savetxt(out, err, fmt='%.6f', delimiter=' ', newline=' ')
out.write('\n')
out.close()

# plot
rscript = open("tmp.r","w")
rscript.write("options(warn=-1)\n")
rscript.write("x <- read.table('"+outfile+"')\n")
rscript.write("pdf('"+outplot+"')\n")
rscript.write("avg <- x[1,]\n")
rscript.write("err <- x[2,]\n")
rscript.write("print(x)\n")
rscript.write("ylim=c(min(avg-err),max(avg+err))\n")
rscript.write("xticks <- seq(ncol(x))\n")
rscript.write("plot(xticks,avg,ylab='average coverage',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.r")
os.system("rm tmp.r")