Population-level structural variant characterization using pangenome graphs - PubMed
a day ago
- #pangenome
- #structural variants
- #deep learning
- Swave is a deep learning-based method for identifying structural variants (SVs) from pangenome graphs.
- It uses 'projection waves' to summarize dotplot images for mapping patterns between reference and SV-indicating alleles.
- A recurrent neural network in Swave distinguishes true SV signals from background noise caused by genomic repeats.
- Swave outperforms existing methods in SV-type classification and genotyping.
- Applied to healthy and rare-disease cohorts, Swave reveals complex and polymorphic SV patterns across human populations.
- Swave identifies potentially pathogenic SVs, aiding in understanding genetic diversity and disease associations.