Filter variants by DP4 (total)?

[Genomics101]

Greetings. I have found that in my particular data set that the most sensitive factor for true variants versus false positives is actually the DP4 (the total number) rather than the QUAL score or depth (DP). Is there a program or little script to filter by the total DP4? Thanks!

[lindenb]

filtering a VCF with javascript

I wrote a tool named vcffilterjs to filter a VCF using a javascript expression: See: https://github.com/lindenb/jvarkit#-...ascript-rhino-

Example:
wth

function accept()
	{
	var DP4=variant.getAttribute("DP4");
	if(DP4==null || DP4.size()!=4) return false;
	return  DP4.get(0)+DP4.get(1)+DP4.get(2)+DP4.get(3)<10;
	}
accept();

$ curl "https://raw.github.com/CBMi-BiG/snpEff/master/tests/vcf_homo.vcf" | \
java -jar dist/vcffilterjs.jar SCRIPT_FILE=filter.js 2> /dev/null

(...)
#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	s_1_ACAGTGA_sort.bam
Y	3720217	.	A	G	8.65	.	AC=2;AF1=1;AN=2;CI95=0.5,1;DP=2;DP4=0,0,0,1;FQ=-30;G3=4.415e-15,5.291e-06,1;MQ=38;SF=5	GT:GQ:PL	0/0:61:60,6,0
Y	3721230	.	C	G	21.80	.	AC=2;AF1=1;AN=2;CI95=0.5,1;DP=2;DP4=0,0,0,2;FQ=-33;G3=1.456e-17,8.564e-07,1;MQ=29;SF=3	GT:GQ:PL	1/1:61:60,6,0
Y	3744605	.	C	A	3.98	.	AC=2;AF1=1;AN=2;CI95=0.5,1;DP=2;DP4=0,0,0,2;FQ=-33;G3=1.468e-15,8.599e-07,1;MQ=19;SF=2	GT:GQ:PL	0/0:61:60,6,0
Y	9945223	.	ATTT	ATTTT	19.80	.	AC=4;AF1=1;AN=4;CI95=0.5,1;DP=2;DP4=0,0,0,2;FQ=-40.5;G3=2.906e-18,8.564e-07,1;INDEL;MQ=45;SF=0,2	GT:GQ:PL	1/1:61:60,6,0

[Genomics101]

@lindenb looks like just what I need, I'll let you know how it goes!