Small variant calling in the era of long-read sequencing.

Deep learning-based tools for small variant calling in long-read sequencing have transformed detection across diverse biological contexts, including germline, de novo, somatic, and mosaic variants.

• Why it matters: Accurate analysis of complex and repetitive genomic regions is essential for understanding human genetic diversity, but long-read sequencing's distinct error profiles require specialized computational approaches.
• What they did: The review discusses the development of deep learning frameworks combined with alignment-based, assembly-based, and pangenome strategies, aiming for complete telomere-to-telomere small variant profiling.
• The result: Advancing these methods will enable benchmarking of complex variants and repetitive regions, facilitating comprehensive genomic analysis and improved variant detection accuracy.