Sharpen SV detection

Lin Tang

DOI: https://doi.org/10.1038/s41592-024-02195-9

IF: 48

2024-02-13

Nature Methods

Abstract:Boosted by the power of long-read sequencing, the frontiers of structural variation (SV) detection are advancing quickly. Accompanying and facilitating this progress, the toolbox of computational methods keeps expanding — although it is still far from being perfected. In 2018, Fritz Sedlazeck at Baylor College of Medicine and colleagues presented 'Sniffles1', which has since become a widely used tool for detecting SV using long-read sequencing data. With new designs, functionalities and improvements, the updated 'Sniffles2' now supports population-scale and somatic SV detection and analysis. A number of innovations contribute to the capacities of Sniffles2. A repeat aware clustering strategy improves SV calling quality, and the generation of genotyped population VCF (variant call format) files facilitates scalable analysis. A new approach to filtering signals from noise enables the detection of mosaic SV, which could be related to different diseases. "Sniffles2 is a complete new implementation led by Moritz Smolka. As such it is significantly faster and more accurate than Sniffles1", says Sedlazeck. Regarding the challenging task of detecting mosaic SV, he says "we now can rely on two or more reads while maintaining a high precision and a good recall on these types of SV". Finally, multiple parts of the code needed to communicate and work together, as noted by Sedlazeck, to merge different concepts and generate genotyped population VCF files across germline and somatic mutations.

biochemical research methods