Apply PASS to vcf Indel file

Question

[hdtms]

In the terminal I ran:

java -jar ../Programs/GenomeAnalysisTK.jar -R ../Programs/human_g1k_v37_decoy.fasta -T SelectVariants --variant Analysis/1.GATK_filtering/1.Only_Indels/vcf_Indels.vcf -select "FILTER == 'PASS'" -o Analysis/1.GATK_filtering/2.Only_Pass/vcf_Indels-Passes.vcf

Which gave me a file with this number of lines

204 vcf_Indels-Passes.vcf

I found this number very low, so I tried doing:

java -jar ../Programs/GenomeAnalysisTK.jar -R ../Programs/human_g1k_v37_decoy.fasta -T SelectVariants --variant Analysis/1.GATK_filtering/1.Only_Indels/vcf_Indels.vcf -select 'vc.isNotFiltered()' -o Analysis/1.GATK_filtering/3.Only_Pass/vcf_Indels-Passes.vcf

Which gave me a file with this number of lines

29531 vcf_Indels-Passes.vcf

I don't know which of these filters is correct or if they're both wrong, next I will be applying statistical test to the resulting PASS file.

Answer 1

[Kevin Blighe]

The first line of code will only extract InDels with a 'PASS' in the VCF FILTER column.

The second line of code may not be functioning as expected. It may be outputting all InDels that either have a PASS or that have no value in the FILTER column. The LowQual ones may still be filtered out. You just have to check the output to see what's happening.

I tend to do my own VCF filtering using awk or Python because, in many situations, the standard tools don't behave as we would expect. This may be due to the fact that the VCF format suffers from a lack of standardisation, generally, i.e., there is a specified format but many variations of it.