Wondering how I can get the number of informative sites in a .vcf file using VCFtools. By informative I mean at least two samples share a variant. Any suggestions? Thanks.
•
written
by
outlier95 •
using vcffilterjs: https://github.com/lindenb/jvarkit/wiki/VCFFilterJS add INFORMATIVE in the FILTER column for the variant having less than two samples having more than one genotype hom-ref or het. extract the FILTER column, count the number of line containing INFORMATIVE
cat input.vcf |\
java -jar dist/vcffilterjs.jar -F INFORMATIVE -e 'function accept(v) { var f=0,i;for(i=0;i<v.getNSamples();++i) {var g=v.getGenotype(i); f+=(g.isHomVar() || g.isHet()?1:0);} return f<2;}accept(variant);' |\
grep -v "^#" | cut -f 7 | grep -c INFORMATIVE
MOD-EDIT: OP has opened a new question for this here: Identifying private and shared SNPs using VCFtools
If I remove singletons and private doubletons via --singletons and --positions using VCFtools, then take the difference in the number of SNPs before and after filtering, should that amount to the number of informative SNPs (per my definition)? Many thanks.
Log in to answer this question.
Not sure about VCFtools, but if you are up for trying something new, the Variant Effect Predictor gives you information on MAF for data in .vcf files. Look at the Filtering options available for the VEP including frequency i.e. MAF.
GATK indicates they do this for their best practices and then makes the reader scavenger around the internet to find how to do this instead of giving a resource. Shame.