how to filter orphan breakends from a vcf?

Question

[joannew]

After several rounds of filtering, I have vcf files in which some of the break-ends are missing their mates (which presumably didn't pass the other filter parameters). Does anyone know of a way to filter out these orphan break-ends? I don't see that there is any flag or annotation within the remaining BND record, just that I have ID# 28061_1, for example, included in the file, but not the mate 28061_2. Thanks for any suggestions!

Answer 1

[Pierre Lindenbaum]

using bioalcidaejdk. http://lindenb.github.io/jvarkit/BioAlcidaeJdk.html

loop over all the BDN, create an associative map ID->MATEID and print the keys where the mate is not found.

java -jar ~/src/jvarkit-git/dist/bioalcidaejdk.jar -e 'final Map<String,String> map=stream().filter(V->V.hasID() && V.getAttributeAsString("SVTYPE","").equals("BND")).collect(Collectors.toMap(V->V.getID(),V->V.getAttributeAsString("MATEID",""))); for(String k:map.keySet()) if(!map.containsKey(map.get(k))) println(k); ' input.vcf > id.list

it gonna print a list of orphan ID that you can use to filter-out the VCF using for example GATK selectvariants and the option --excludeIDs using https://software.broadinstitute.org/gatk/documentation/tooldocs/3.8-0/org_broadinstitute_gatk_tools_walkers_variantutils_SelectVariants.php#--excludeIDs