K-mer analysis of long-read alignment pileups for structural variant genotyping

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Accurate genotyping of structural variants (SVs) is essential for advancing genomics research.
• Existing methods for SV genotyping face challenges in accuracy and efficiency, particularly in complex genomic regions.

Purpose of the Study:

• To introduce kanpig, a novel computational method for rapid and accurate SV genotyping.
• To evaluate kanpig's performance against current state-of-the-art SV genotyping tools.

Main Methods:

• Kanpig utilizes variant graphs and k-mer vectors for SV genotyping.
• The method was benchmarked on single-sample and multi-sample datasets using long-read sequencing data (PacBio and ONT).

Main Results:

• Kanpig achieved a single-sample genotyping concordance of 82.1%, significantly outperforming existing tools (average 66.3%).
• The method demonstrated high accuracy in multi-sample projects, especially at complex loci with neighboring SVs.
• Processing a single sample's 20x long-reads takes only 43 seconds.

Conclusions:

• Kanpig offers a substantial improvement in SV genotyping accuracy and efficiency.
• The method is applicable to both single-sample and large-scale multi-sample genomics projects.
• Kanpig provides a valuable new tool for researchers studying structural variations in diverse populations.