Skip to main content

Table 4 Overall accuracy of the SVM classifier trained using the gene panel proposed by OncoPaD and the genes selected in this study via SVM-RFE and stability selection

From: Predicting cancer type from tumour DNA signatures

Classification task

277 OncoPaD tier 1 genes

Top 277 SVM-RFE-based SPM genes

Top 277 SVM-RFE-based SPM and CNA genes

28 cancer types of this study

49.6 %

57.3 %

88.1 %

19 cancer types common between this study and OncoPaD

56.0 %

63.4 %

90.3 %

  1. The 19 tumour types that are common to our data set and OncoPaD are those labelled 1–11, 13, 14, and 16–21 in Table 1
  2. CNA copy number altered, SPM somatic point-mutated, SVM support vector machine, SVM-RFE SVM recursive feature elimination