Hilary Elrick from EMBL-EBI in the UK spoke at London Calling 2023 about “SAVANA: a computational method to characterise structural variation in human cancer genomes using nanopore sequencing.” They spoke about the structural variants in cancer and how much is still unknown or understudied. Elrick studied sarcoma and glioblastoma. Sarcomas are cancers of bone and soft-tissue and understudied. There are fifty different subtypes, yet Elrick noted that little has changed in treatment outcomes since the 1980s. Glioblastoma is the most common type of brain tumour and is aggressive and treatment-resistant. Elrick and collaborators sequenced 21 sarcomas and 14 glioblastomas with 39x and 28x mean normal coverage, respectively. Elrick developed SAVANA to analyze tumor and non-tumor BAM files. A machine learning algorithm was used to detect Structural variants. They re-sequenced the COLO829 cells with R10 and used for analysis. Elrick and team compared SAVANA to other callers and used an R package developed by the lab to create figures to depict coverage, breakpoints, and coverage. SAVANA is available and being improved. Additional updates may improve the classifier and SV calling/representation. Elrick shared several examples of benchmarking and comparisons to other callers, highlighting how it is a great tool to use for cancer genomics.
