Minor allele frequency thresholds strongly affect population structure inference with genomic data sets.

Area of Science:

• Population genetics
• Genomics
• Bioinformatics

Background:

• Minimizing errors in large DNA sequence datasets often involves filtering sites with low minor allele frequency (MAF).
• The impact of MAF filtering on downstream population genetic inferences is not well understood.
• Population structure analysis is a critical first step in population genomics.

Purpose of the Study:

• To investigate the effects of MAF thresholds on population structure inference.
• To demonstrate how MAF filtering can alter population genetic analyses.
• To provide recommendations for applying MAF filters in population genomics.

Main Methods:

• Simulations were used to model the impact of MAF thresholds.
• An empirical single nucleotide polymorphism (SNP) dataset was analyzed.
• Model-based and multivariate methods were employed to infer population structure.

Main Results:

• Including singletons (variants with MAF below a threshold) confounds population structure inference.
• More stringent MAF cutoffs lead to less distinct population clusters in both model-based and multivariate analyses.
• The observed effects are attributed to data matrix size reduction and allele frequency-mutational age correlations.

Conclusions:

• MAF filtering significantly impacts population structure inference in genomic data.
• Careful consideration of MAF thresholds is necessary to avoid biased population genetic results.
• Best practices for MAF filtering are recommended for studies of population structure using genomic data.