Pan-cancer detection of driver genes at the single-patient resolution

Abstract

Background Identifying the complete repertoire of genes that drive cancer in individual patients is crucial for precision oncology. Most established methods identify driver genes that are recurrently altered across patient cohorts. However, mapping these genes back to patients leaves a sizeable fraction with few or no drivers, hindering our understanding of cancer mechanisms and limiting the choice of therapeutic interventions. Results We present sysSVM2, a machine learning software that integrates cancer genetic alterations with gene systems-level properties to predict drivers in individual patients. Using simulated pan-cancer data, we optimise sysSVM2 for application to any cancer type. We benchmark its performance on real cancer data and validate its applicability to a rare cancer type with few known driver genes. We show that drivers predicted by sysSVM2 have a low false-positive rate, are stable and disrupt well-known cancer-related pathways. Conclusions sysSVM2 can be used to identify driver alterations in patients lacking sufficient canonical drivers or belonging to rare cancer types for which assembling a large enough cohort is challenging, furthering the goals of precision oncology. As resources for the community, we provide the code to implement sysSVM2 and the pre-trained models in all TCGA cancer types (https://github.com/ciccalab/sysSVM2).

Comparative assessment of genes driving cancer and somatic evolution in noncancer tissues: an update of the NCG resource

“…This gap can be solved by complementing cohortlevel approaches with methods that account for all types of alterations and predict drivers in individual samples, for example identifying their network deregulations 64,65,66 or applying machine learning to identify driver alterations 67 . Alternatively, we have shown that systems-level properties capture the main features of cancer drivers, justifying their use for patient-level driver detection 68,69 .…”

Molecular-based precision oncology clinical decision making augmented by artificial intelligence

“…More recently, PANOPLY incorporated clinical features in addition to omics data and applied random forest analysis to identify prioritized treatment given a patient's clinical and molecular profile ( http://kalarikrlab.org/Software/Panoply.html ) and some anecdotal success was reported [ 36 ]. Nulsen et al [ 37 ] developed a one-class support vector machine called sysSVM that was trained on TCGA data to create a pan-cancer detection tool for identifying driver genes at the granularity of single patients ( https://github.com/ciccalab/sysSVM2 ). Computational validation has shown promising results in terms of low false positive rate which is very essential for the clinical utilization but none of the applications has been formally evaluated in clinical contexts.…”

Machine learning methods for prediction of cancer driver genes: a survey paper

“…Moreover, integrating different data categories is not necessarily a recent tendency, as papers published in 2009 and 2011 already proposed such strategy. Nonetheless, most of the papers using features from three or more data categories were published from 2015, and those that used four [34,57] or five [54,59] data categories were published mainly in 2020 and 2021, which may indicate a trend for increasing data diversity in newer models.…”

“…Davoli et al [23] proposed an entropy-based mutation selection score that reflects the spatial distribution of these features, measuring the preferred occurrence of specific point mutations within a gene, termed 'mutation hotspots'. This measure of positional clustering was further adopted by several papers [29,34,45,48,54,59]. Mutation hotspots were also detected using scores computed by OncoDriveCLUST [26,59,64] and applying density estimates to aggregate closelyspaced missense mutations into peaks and compute mutation fraction inside the highest peak [53].…”

“…This measure of positional clustering was further adopted by several papers [29,34,45,48,54,59]. Mutation hotspots were also detected using scores computed by OncoDriveCLUST [26,59,64] and applying density estimates to aggregate closelyspaced missense mutations into peaks and compute mutation fraction inside the highest peak [53]. The normalized number of SNPs in the exon where the mutation is located [24], mutations' distance to closest Transcribed Sequence Start (TSS) and closest Transcribed Sequence End (TSE) [43], and a genelevel binary matrix summarizing mutation occurrence across all samples [33,51] were also adopted.…”