sysSVM2

Pan-cancer detection of driver genes at the single-patient resolution. sysSMV2 is a computational tool for patient-specific cancer driver gene prioritisation. It is based on the principle that driver genes are characterised by particular molecular properties (e.g. mutations, copy number variants) and systems-level properties (e.g. evolutionary origin, breadth of expression). It works by identifying genes with similar properties to canonical drivers, using a one-class Support Vector Machine framework1. Users may also train their own sysSVM2 models. Models that have been trained on data from TCGA are available to download from the appropriate directory. For smaller TCGA cohorts (N <200 samples), we recommend using the model trained on pan-cancer data. To download sysSVM2, clone this repository. sysSVM2 is implemented in R. In this guide, we assume that the user's working directory corresponds to a clone of this repository.