Evaluating genotype imputation pipeline for ultra-low coverage ancient genomes

Area of Science:

• Genetics
• Paleogenomics
• Bioinformatics

Background:

• Ancient DNA (aDNA) analysis is crucial for understanding past populations.
• High-coverage genomes are difficult to obtain from degraded aDNA samples.
• Low-coverage aDNA limits downstream population genetic analyses.

Purpose of the Study:

• To evaluate a two-step imputation pipeline for common variants in ancient genomes with ultra-low coverage (0.05-1x).
• To assess the accuracy and effectiveness of imputation for improving the usability of ancient samples.

Main Methods:

• Developed and tested a two-step imputation pipeline using Beagle genotype likelihoods and confident genotype filtering.
• Compared imputation accuracy across various genotype calling methods, filters, reference panels, and imputation tools.
• Evaluated imputation performance on a Neolithic Hungarian genome at different coverage levels.

Main Results:

• The two-step pipeline achieved ~90% imputation accuracy for heterozygous common variants at 0.05x coverage and >97% at 0.5x coverage.
• This approach outperformed single-step imputation, indicating current genotype callers may not fully address ancient DNA errors.
• Imputation successfully mitigated, but did not eliminate, reference bias in ultra-low coverage ancient genomes.

Conclusions:

• A two-step imputation strategy enhances genotype accuracy in low-coverage ancient genomes.
• This pipeline improves the utility of degraded ancient DNA samples for population genetic research.
• Further quality control steps in imputation pipelines are beneficial for ancient DNA data.