Automating Variant Filtering with VCFTools: Step-by-Step Examples

Variant Call Format (VCF) files are central to genome analysis pipelines. VCFTools is a widely used suite for filtering, summarizing, and manipulating VCFs. This article shows practical, repeatable steps to automate variant filtering with VCFTools using command-line examples and short scripts so you can integrate them into pipelines.

Prerequisites

VCFTools installed (vcftools CLI). Install via package manager or from source.
Input VCF (compressed .vcf.gz recommended) and its index (.tbi) if using bgzip/Tabix.
Basic shell (bash) familiarity.

Goals and assumptions

Remove low-quality variants.
Exclude variants with low call rate (high missingness).
Filter by minor allele frequency (MAF).
Separate SNPs and indels.
Produce a final compressed VCF ready for downstream tools.

Assumptions: input file is sample.vcf.gz. Adjust filenames and thresholds as needed.

Step 1 — Prepare files

Ensure VCF is bgzipped and indexed (Tabix).

bash
bgzip -c sample.vcf > sample.vcf.gz tabix -p vcf sample.vcf.gz

Step 2 — Basic quality filtering

Filter by phred-scaled quality (QUAL) and genotype quality (GQ). VCFTools operates on site- and genotype-level filters; use –minQ for site QUAL, and –minGQ for genotype GQ.

bash
vcftools –gzvcf sample.vcf.gz 
–minQ 30 
         –minGQ 20 
         –recode –recode-INFO-all 
         –out filtered_q bgzip -c filtered_q.recode.vcf > filteredq.recode.vcf.gz

Step 3 — Filter by missingness (call rate)

Remove sites with high missing data; e.g., keep sites with at least 90% call rate (–max-missing 0.9).

bash
vcftools –gzvcf filtered_q.recode.vcf.gz –max-missing 0.9 –recode –recode-INFO-all –out filtered_q_miss bgzip -c filtered_q_miss.recode.vcf > filtered_qmiss.recode.vcf.gz

Step 4 — Filter by minor allele frequency (MAF)

Exclude rare variants below desired frequency, e.g., MAF < 0.01:

bash
vcftools –gzvcf filtered_q_miss.recode.vcf.gz –maf 0.01 –recode –recode-INFO-all –out filtered_q_miss_maf bgzip -c filtered_q_miss_maf.recode.vcf > filtered_q_missmaf.recode.vcf.gz

Step 5 — Separate SNPs and indels

VCFTools can extract SNPs using –remove-indels or extract indels with –keep-only-indels.

bash
# SNPs only vcftools –gzvcf filtered_q_miss_maf.recode.vcf.gz –remove-indels –recode –recode-INFO-all –out final_snps bgzip -c final_snps.recode.vcf > final_snps.recode.vcf.gz # Indels only vcftools –gzvcf filtered_q_miss_maf.recode.vcf.gz –keep-only-indels –recode –recode-INFO-all –out final_indels bgzip -c final_indels.recode.vcf > finalindels.recode.vcf.gz

Step 6 — Filtering by depth (DP)

Filter sites by depth (e.g., min-meanDP 10, max-meanDP 200).

bash
vcftools –gzvcf final_snps.recode.vcf.gz –min-meanDP 10 –max-meanDP 200 –recode –recode-INFO-all –out final_snps_dp bgzip -c final_snps_dp.recode.vcf > final_snpsdp.recode.vcf.gz

Step 7 — Combine filters in one run (example)

You can combine many filters in a single vcftools invocation to speed up processing:

bash
vcftools –gzvcf sample.vcf.gz –minQ 30 –minGQ 20 –max-missing 0.9 –maf 0.01 –remove-indels –min-meanDP 10 –max-meanDP 200 –recode –recode-INFO-all –out combined_filtered_snps bgzip -c combined_filtered_snps.recode.vcf > combined_filteredsnps.recode.vcf.gz

Step 8 — Automate with a shell script

Create a reusable script that takes input VCF and parameters.

bash
#!/usr/bin/env bash set -euo pipefail IN_VCF=${1:-sample.vcf.gz} OUT_PREFIX=${2:-filtered} MINQ=${3:-30} MINGQ=${4:-20} MAXMISS=${5:-0.9} MAF=${6:-0.01} MINDP=${7:-10} MAXDP=${8:-200} vcftools –gzvcf “${IN_VCF}" --minQ "${MINQ}” –minGQ “${MINGQ}" --max-missing "${MAXMISS}” –maf “${MAF}" --remove-indels --min-meanDP "${MINDP}” –max-meanDP “${MAXDP}" --recode --recode-INFO-all --out "${OUT_PREFIX}_snps” bgzip -c “${OUT_PREFIX}_snps.recode.vcf" > "${OUT_PREFIX}_snps.recode.vcf.gz” tabix -p vcf “${OUT_PREFIX}snps.recode.vcf.gz”

Make executable and run:

bash
chmod +x filter_vcftools.sh ./filtervcftools.sh sample.vcf.gz myfiltered 30 20 0.9 0.01 10 200

Step 9 — Reporting and QC

VCFTools can produce summary statistics useful for QC:

bash
vcftools –gzvcf sample.vcf.gz –TsTv-summary –out ts_tv vcftools –gzvcf sample.vcf.gz –depth –out depth_stats vcftools –gzvcf sample.vcf.gz –missing-site –out missing_site

Inspect outputs (text files) or parse them into plots with R/Python.

Tips and best practices

Keep intermediate files for reproducibility or use a workflow manager (Snakemake/Nextflow).

Tune thresholds to cohort size and sequencing platform.

For per-sample filters (e.g., sample missingness), use –missing-indv and remove samples with high missingness.

Consider complementary tools (bcftools, GATK) for complex filters or annotation.

Example pipeline outline

bgzip + tabix input

Combined vcftools filtering (quality, missingness, MAF, depth)

Separate SNPs/indels

Index outputs

Run QC summaries

Archive filtered VCFs

This sequence gives a reproducible, automatable approach to variant filtering with VCFTools. Adjust flags and thresholds to match your study design and sequencing characteristics.

Automating Variant Filtering with VCFTools: Step-by-Step Examples