Genetic Variant Filtering Guide

Welcome! This page breaks down the essential filters used for the refinement of genetic variant data before the downstream analysis. From removing errors to keeping only high-confidence SNPs, each step here helps you build more accurate dataset for things like population genetics, GWAS, and PCA. Whether you're new to variant filtering or just need an in-depth understanding of the step, this guide makes complex filters easy to understand and even easier to apply. The table shows the names we use to refer to the filters we apply.

Filter Name	Description
`passOnly`	Retained only variants that passed all internal quality filters from the variant caller (Strelka in this case).
`biallelicOnly`	Kept only biallelic variants (i.e., sites with exactly two alleles) for downstream compatibility and clarity.
`rmvIndels`	Removed insertions and deletions (INDELs), keeping only SNPs (single nucleotide polymorphisms).
`minMAF0Pt05`	Filtered for variants with a minor allele frequency (MAF) ≥ 0.05, ensuring informative polymorphisms and removing rare variants.
`chr_E2`	Restricted to a specific chromosome/region of interest (e.g., `chr_E2`).
`minDP3`	Required a minimum depth (DP) of 3 per genotype to avoid false positives due to low coverage.
`minQ30`	Ensured that each site has a minimum site quality score of 30, reflecting high confidence in the variant.
`minGQ30`	Filtered genotypes to retain only those with a Genotype Quality (GQ) ≥ 30, removing uncertain genotype calls.
`hwe_0.05`	Removed variants deviating significantly from Hardy-Weinberg Equilibrium (p < 0.05).
`imiss_0.6`	Filtered out individuals with >60% missing genotypes to maintain data quality.
`miss_0.6`	Removed variants missing in >60% of individuals, ensuring sufficient representation across samples.
`mid95percentile`	Retained the middle 95% of data based on depth distribution, trimming extreme outliers.
`noZSB`	Custom filter

Minor Allele Frequency (MAF)

MAF is the frequency at which the less common allele occurs at a genetic locus in a population.

Example: If allele A appears in 90% of individuals and G in 10%, then MAF = 0.10.

Why apply `minMAF0Pt05`?

Keeps informative, common variants
Reduces noise from rare alleles or sequencing errors
Improves reliability for analyses like PCA, GWAS, or Fst

Depth Filter: `minDP3`

This filter ensures that each genotype (or site) is supported by at least 3 reads. Variants with DP < 3 are excluded.

Ensures reliable variant calls
Reduces errors from low-coverage or spurious reads
Improves accuracy in downstream analyses

Quality Filter: `minQ30`

This filter retains only variants with a site quality score ≥ 30, indicating high confidence in the variant call. QUAL 30 = 99.9% confidence the variant is true.

Removes low-quality or uncertain variants
Reduces false positives
Improves overall reliability of the dataset

Genotype Quality Filter: `minGQ30`

The genotype quality (GQ) score represents the confidence in the individual genotype call at a variant site. GQ 30 = 99.9% confidence that the genotype is correct.

Removes low-confidence genotype calls
Reduces false genotyping results
Improves the accuracy of downstream sample-level analyses

Hardy-Weinberg Equilibrium (HWE)

The filter hwe_0.05 removes variants that deviate significantly from HWE (p-value < 0.05).

May indicate genotyping errors, population structure, or selection
Ensures biological validity in population-level studies

Missingness per Sample: `imiss_0.6`

Removes samples with more than 60% missing genotype data.

Removes highly incomplete samples that can bias PCA or ADMIXTURE
Balances data quality and retention
Improves dataset reliability

Missingness per Site: `miss_0.6`

Removes variant sites where more than 60% of samples have missing data.

Ensures you're analyzing variants seen in most samples
Reduces noise from poorly genotyped regions

What is `mid95percentile`?

This filter retains only those variants where the depth per site falls within the middle 95% of the overall depth distribution.

Removes extreme outliers: Cuts the lowest 2.5% and highest 2.5%
Retains most of the data: 95% window preserves the bulk of biologically informative variants
Balances stringency and power: Avoids skew from extreme values

Genetic Variant Filtering Guide

Minor Allele Frequency (MAF)

Why apply minMAF0Pt05?

Depth Filter: minDP3

Quality Filter: minQ30

Genotype Quality Filter: minGQ30