De novo and somatic structural variant discovery with SVision-pro
Songbo Wang, Jiadong Lin, Peng Jia, Tun Xu, Xiujuan Li, Y Liu, Dan Xu, Stephen J. Bush, Deyu Meng, Kai Ye
Abstract
Long-read-based de novo and somatic structural variant (SV) discovery remains challenging, necessitating genomic comparison between samples. We developed SVision-pro, a neural-network-based instance segmentation framework that represents genome-to-genome-level sequencing differences visually and discovers SV comparatively between genomes without any prerequisite for inference models. SVision-pro outperforms state-of-the-art approaches, in particular, the resolving of complex SVs is improved, with low Mendelian error rates, high sensitivity of low-frequency SVs and reduced false-positive rates compared with SV merging approaches.
Topics & Concepts
GenomeComputational biologyBiologySomatic cellInferenceFalse discovery rateSegmentationStructural variationGeneticsArtificial intelligenceComputer scienceGeneGenomics and Phylogenetic StudiesChromosomal and Genetic VariationsGenetic diversity and population structure