Personalized pangenome references

Area of Science:

• Genomics
• Bioinformatics

Background:

• Pangenomes offer improved genetic diversity representation over single references.
• Comparing samples to pangenomes can introduce errors from irrelevant variants, often filtered by allele frequency, which is suboptimal.

Purpose of the Study:

• To develop a novel approach for accurate variant genotyping within pangenome graphs.
• To address the challenge of irrelevant variants in pangenome comparisons.

Main Methods:

• Imputation of a personalized pangenome subgraph by sampling local haplotypes.
• Utilizing k-mer counts from sequencing reads for imputation.
• Implementation within the vg toolkit for the Giraffe short-read aligner.

Main Results:

• Reduced small variant genotyping errors by fourfold compared to Genome Analysis Toolkit.
• Achieved short-read structural variant genotyping accuracy competitive with long-read methods.

Conclusions:

• The proposed method enhances the accuracy of pangenome-based variant calling.
• This approach improves the utility of pangenomes for diverse genomic analyses.