Modeling Biases from Low-Pass Genome Sequencing to Enable Accurate Population Genetic Inferences

Area of Science:

• Population Genomics
• Bioinformatics
• Computational Biology

Background:

• Low-pass genome sequencing offers a cost-effective way to study large populations.
• However, it introduces biases, particularly affecting heterozygous genotypes and low-frequency alleles, which complicates demographic history inference.
• Existing methods to correct allele frequency spectrum (AFS) biases can introduce additional noise.

Purpose of the Study:

• To develop a novel approach for demographic inference directly from low-pass sequencing data.
• To incorporate the specific biases of low-pass data into population genetic models, avoiding AFS correction.
• To improve the accuracy of demographic parameter estimation from cost-effective, large-scale genomic datasets.

Main Methods:

• Developed a probabilistic model that accounts for biases introduced by the Genome Analysis Toolkit (GATK) multisample calling pipeline.
• Integrated this model into the dadi software for population genomic inference.
• Evaluated the model using simulated low-pass datasets and by downsampling data from the 1000 Genomes Project.

Main Results:

• The new model effectively alleviates biases present in low-pass sequencing data.
• Inferred demographic parameters showed improved accuracy compared to methods relying on AFS correction.
• Validation on real-world data confirmed the model's practical applicability and effectiveness.

Conclusions:

• This method provides a robust way to perform demographic inference directly from low-pass sequencing data.
• It significantly enhances the reliability of population genomic studies using cost-effective sequencing strategies.
• The approach is broadly applicable, improving inferences for large cohort genomic analyses.