view align2database.py @ 11:b7f1d9f8f3bc

Uploaded

'''
align mulitple bed to one bed
python align_multiple.py hmChIP_mm9_peak_bed/mm9-GSE19999_PolII_P25_all.cod.bed hmChIP_mm9_peak_bed/ test.txt test.pdf 100 5000
'''

import os,sys,random
def main():
    queryfile = sys.argv[1]
    inpath = sys.argv[2]
    outputdata = sys.argv[3]
    outputerr = sys.argv[4]
    barplotpdf = sys.argv[5]
    min_feat = sys.argv[6] # min features overlap
    windowsize = sys.argv[7]
    anchor = sys.argv[8]
    span = sys.argv[9] # loess smooth parameter
        
    inpath = inpath.rstrip('/')
    #os.system('rm '+inpath+'/*tmp*')

    infiles = os.listdir(inpath)

    #print len(infiles),' files\n'
    i = 0
    for infile in infiles:
        if 'tmp' in infile:
            #os.system('rm '+inpath+'/'+infile)
            continue
        i = i +1
        print i,infile
        output = infile.split('/')[-1]+'-to-'+queryfile.split('/')[-1]#'.summary'
        if anchor == 'database':
            command = 'python /Users/xuebing/galaxy-dist/tools/mytools/alignr.py -b '+inpath+'/'+infile+' -a '+queryfile+' -o '+output+' --summary-only -q -w '+windowsize
        else:
            command = 'python /Users/xuebing/galaxy-dist/tools/mytools/alignr.py -a '+inpath+'/'+infile+' -b '+queryfile+' -o '+output+' --summary-only -q -w '+windowsize            
        #print command+'\n'
        os.system(command)
    print 'start visualization...'
    # visualize
    rscriptfile = 'f-'+str(random.random())+'.r'
    r = open(rscriptfile,'w')
    r.write("files <- dir('.','summary',full.name=T)\n")
    #r.write("print(files)\n")    
    r.write("x <- read.table(files[1])\n")
    r.write("err <- read.table(gsub('summary','standarderror',files[1]))\n")
    r.write("for (filename in files[2:length(files)]){\n")
    r.write("   x <- rbind(x,read.table(filename))\n")
    r.write("   err <- rbind(err,read.table(gsub('summary','standarderror',filename)))\n")    
    r.write("}\n")
    r.write("x <- x[x[,2] > "+min_feat+",]\n")
    r.write("err <- err[err[,2] > "+min_feat+",]\n")    
    r.write("name <- as.character(x[,1])\n")
    r.write("nfeat <- x[,2]\n")
    r.write("x <- x[,3:ncol(x)]\n")
    r.write("err <- err[,3:ncol(err)]\n")    
    r.write("for (i in 1:nrow(x)) {\n")
    r.write("    name[i] <- strsplit(name[i],'-to-')[[1]][1]\n")
    r.write("}\n")
    # remove rows that have no variation, which cause problem in heatmap. This is the case when align to itself
    r.write("toremove <- seq(nrow(x))\n")
    r.write("for (i in 1:nrow(x)){\n")
    r.write("    toremove[i] <- all(x[i,] == x[i,1])\n")
    r.write("}\n")
    r.write("x <- x[!toremove,]\n")
    r.write("err <- err[!toremove,]\n")
    r.write("name <- name[!toremove]\n")
    r.write("nfeat <- nfeat[!toremove]\n")
    r.write("write.table(cbind(name,nfeat,x),file='"+outputdata+"',sep ='\\t',quote=F,row.names=F,col.names=F)\n")
    r.write("write.table(cbind(name,nfeat,err),file='"+outputerr+"',sep ='\\t',quote=F,row.names=F,col.names=F)\n")
        
    r.write("pdf('"+barplotpdf+"')\n")
    r.write("heatmap(t(scale(t(as.matrix(x,nc=ncol(x))))),Colv=NA,labRow=name,labCol=NA,margins=c(1,8),cexRow=0.02+1/log(nrow(x)),col=heat.colors(1000))\n")

    if windowsize != '0' :
        r.write("xticks <- seq(-"+windowsize+","+windowsize+",length.out=100)\n")
        r.write("xlab <- 'relative position (bp)'\n")
    else:
        r.write("xticks <- seq(100)\n")
        r.write("xlab <- 'relative position (bin)'\n")
        
    #r.write("plot(xticks,colSums(t(scale(t(as.matrix(x,nc=ncol(x)))))),xlab='relative position (bp)',type='l',lwd=2,main='total signal')\n")
    r.write("for (i in 1:nrow(x)) {\n")
    r.write("   avg <- x[i,]/nfeat[i]\n")
    #r.write("   maxv <- max(avg)\n")
    #r.write("   minv <- min(avg)\n")
    #r.write("   medv <- median(avg)\n")
    #r.write("   if (maxv < "+fold+"*medv | minv*"+fold+">medv){next}\n")
    #smooth
    if float(span) >= 0.1:
        r.write("   avg = loess(as.numeric(avg)~xticks,span="+span+")$fitted\n")
        r.write("   err[i,] = loess(as.numeric(err[i,])~xticks,span="+span+")$fitted\n")
    r.write("   par(cex=1.5)\n")
    r.write("   plot(xticks,avg,ylab='average coverage',main=paste(name[i],'\n n=',nfeat[i],sep=''),xlab=xlab,type='l',lwd=1,ylim=c(min(avg-err[i,]),max(avg+err[i,])))\n")   
    r.write("   polygon(c(xticks,rev(xticks)),c(avg+err[i,],rev(avg-err[i,])),col='slateblue1',border=NA)\n")
    r.write("   lines(xticks,avg,type='l',lwd=1)\n")   
    r.write("}\n")
    r.write("dev.off()\n")
    r.close()
    os.system("R --vanilla < "+rscriptfile)    
    os.system('rm '+rscriptfile)
    os.system('rm *.summary')
    os.system('rm *.standarderror')

main()