How to merge multiple CNV results of CNVnator

Question

[ChrisJ]

Hello everyone: There is a problem puzzled me recently,I have got every individual CNV output by the tool of CNVnator,and next I want to merge all outputs to a file(if two CNVs have overlap >0.5 can be considered to a common one), but I don't know how to merge these files.I have tried the VCFtools( CNVnator has a perl script which can switch the out to VCF format),bcftools,bedtools ,but there is no a tool can really solve my question(some tool would discard these no-overlap CNVs),so who can give me some suggestions to solve this hard trouble? this is my output:

  # chr       start            end           type
  chr1    7967601     7984800     deletion
  chr1    9441201     9447200     duplication
  chr1    10546001   10553200    deletion
  chr1    10822001    10830400   deletion

[hzaujsxu]

I met the same problem. Have you solve it?

[Jingjingzhang]

hi, I also met the same problem ...

Answer 1

[Pierre Lindenbaum]

a 'stupid' solution using javascript ?

// following array is generated with awk '{printf("{\"chrom\":\"%s\",\"start\":%s,\"end\":%s,\"type\":\"%s\"},\n",$1,$2,$3,$4);}' your.bed
var segments=[
{"chrom":"chr1","start":7967601,"end":7984800,"type":"duplication"},
{"chrom":"chr1","start":7967602,"end":7984810,"type":"duplication"},
{"chrom":"chr1","start":7967602,"end":7967603,"type":"duplication"},
{"chrom":"chr1","start":1,"end":100,"type":"duplication"},
{"chrom":"chr1","start":10546001,"end":10553200,"type":"deletion"},
{"chrom":"chr1","start":10546101,"end":10553000,"type":"deletion"}
]

var done=false;
while(done==false)
    {
    done=true;
    var i=0;
    while(done && i + 1< segments.length)
        {
        var j = i+1;
        while(j < segments.length)
            {
            var segi = segments[i];
            var segj = segments[j];
            if( segi.chrom != segj.chrom) {++j; continue;}
            if( segi.type != segj.type) {++j; continue;}
            if( segj.end <= segi.start)  {++j; continue;}
            if( segj.start >= segi.end)  {++j; continue;}
            var overlap_len = 1.0 * ( Math.min(segi.end,segj.end) - Math.max(segi.start,segj.start) );
            var leni =  1.0* (segi.end - segi.start);
            if( overlap_len / leni < 0.5)  {++j; continue;}
            var lenj =  1.0* (segj.end - segj.start);
            if( overlap_len / lenj < 0.5)  {++j; continue;}
            segments.splice(j,1);
            segments[i]={
                "chrom":segi.chrom,
                "start": Math.min(segi.start,segj.start),
                "end": Math.max(segi.end,segj.end),
                "type":segi.type
                }
            done=false;
            }
        i++;
        }
    }

for(var i in segments)
    {
    var segi = segments[i];
    print(segi.chrom+"\t"+segi.start+"\t"+segi.end+"\t"+segi.type);
    }

invoke with jjs or in a firefox scratchad

$ jjs script.js
chr1    7967601 7984810 duplication
chr1    7967602 7967603 duplication
chr1    1   100 duplication
chr1    10546001    10553200    deletion

Answer 2

[ChrisJ]

ok,thank you very much .and I will try it .