Accurate somatic small variant discovery for multiple sequencing technologies with DeepSomatic
Cancer is a highly heterogeneous disease, with thousands of somatic variants across different cells, requiring accurate characterisation. Typically, cancer research relies on short-read sequencing, but limitations of this approach include the inability to resolve repetitive or complex regions. With Oxford Nanopore sequencing, any length of DNA can be sequenced and analysed, revealing areas that were previously inaccessible. Here, the authors present DeepSomatic, a somatic small variant caller that combines long- and short-read data to accurately detect somatic small nucleotide variants and indels. They found that this bioinformatic tool outperformed other somatic variant callers, achieving consistently high F1-scores.
‘DeepSomatic’s models specific to FFPE and WES also prove to outperform other variant callers that were not specifically trained for these data types’
Park and Cook et al. Nature Biotechnology (2025)
Sample type: glioblastoma tumour tissue and matched peripheral blood mononuclear cells
