StableLift: Optimized Germline and Somatic Variant Detection Across Genome Builds

Area of Science:

• Genomics
• Bioinformatics

Background:

• Reference genomes are essential for genomic analysis but evolve, leading to new builds with incompatible coordinate systems.
• This evolution necessitates understanding the impact of genome build changes on variant calling.

Purpose of the Study:

• To quantify the effect of different human genome builds on germline and somatic variant calling.
• To develop a method for predicting the stability of variants across genome builds.

Main Methods:

• Analysis of tumor-normal whole-genome pairs against two major human genome builds.
• Development and validation of the StableLift algorithm to predict cross-build variant stability.

Main Results:

• Significant build-discordance observed: 3.8% for germline SNPs, 8.6% for germline SVs, 25.9% for somatic SNVs, and 49.6% for somatic SVs.
• 47% of build-discordant variants were validated by targeted resequencing, indicating they are not solely false positives.
• StableLift achieved high predictive performance (AUROC 0.934 ± 0.029) for cross-build variant stability.

Conclusions:

• Genome build differences introduce substantial artifacts in variant calling, requiring caution in cross-build analyses.
• StableLift offers an efficient computational solution to mitigate inter-build artifacts and improve genomic data reliability.