Comprehensive assessment of whole-genome sequencing-based variant calling pipelines in monozygotic twins

Abstract

Whole-genome sequencing (WGS) has emerged as a cornerstone technology in precision medicine, genetic diagnostics, and forensic identification. Compared to traditional capillary electrophoresis, WGS captures the entire genomic landscape and thus support high-resolution individual identification and even the differentiation of monozygotic (MZ) twins with nearly identical DNA contents. The minute genetic variation between MZ twins offer a rigorous paradigm for assessing the accuracy and sensitivity of different variant calling algorithms. Therefore, we benchmarked eight state-of-the-art variant calling pipelines using MZ twin genomes sequenced on the Illumina NovaSeq 6000 platform at 30× and 60× coverage. The precision, robustness, efficiency, and concordance of these pipelines were systematically evaluated against a polymerase chain reaction-validated gold standard set of small variants for MZ twin discrimination. Our results indicated that DRAGEN read aligner combined with BCFtools achieved the optimal balance between accuracy and computationally efficiency, albeit with minor specificity and robustness tradeoffs. GATK HaplotypeCaller exhibited the highest accuracy and sensitivity among all tested pipelines, but required exceptionally high computational resources. DRAGEN and Strelka2 pipeline showcased superior robustness with the highest true-positive ratio and the lowest false-positive ratio. Collectively, these insights gained from the detection of subtle variations within highly similar genomes provide critical guidelines applicable to scenarios demanding high-precision variant calling.