FI-Net: Identification of Cancer Driver Genes by Using Functional Impact Prediction Neural Network

Abstract

Identification of driver genes, whose mutations cause the development of tumors, is crucial for the improvement of cancer research and precision medicine. To overcome the problem that the traditional frequency-based methods cannot detect lowly recurrently mutated driver genes, researchers have focused on the functional impact of gene mutations and proposed the function-based methods. However, most of the function-based methods estimate the distribution of the null model through the non-parametric method, which is sensitive to sample size. Besides, such methods could probably lead to underselection or overselection results. In this study, we proposed a method to identify driver genes by using functional impact prediction neural network (FI-net). An artificial neural network as a parametric model was constructed to estimate the functional impact scores for genes, in which multi-omics features were used as the multivariate inputs. Then the estimation of the background distribution and the identification of driver genes were conducted in each cluster obtained by the hierarchical clustering algorithm. We applied FI-net and other 22 state-of-the-art methods to 31 datasets from The Cancer Genome Atlas project. According to the comprehensive evaluation criterion, FI-net was powerful among various datasets and outperformed the other methods in terms of the overlap fraction with Cancer Gene Census and Network of Cancer Genes database, and the consensus in predictions among methods. Furthermore, the results illustrated that FI-net can identify known and potential novel driver genes.

DriverGenePathway: Identifying driver genes and driver pathways in cancer based on MutSigCV and statistical methods

“…For the performance analysis of DriverGene, parameters categ _ flag , bmr , output _ filestem , p _ class , and sigThreshold were set as defaults.The minimal core driver genes are defined as the genes that are statistically significant in all five methods of hypothesis testing ( q ≤ sigThreshold ). According to the previous studies [27] , [28] , the validity of DriverGene was evaluated by the overlap rate between the identified driver genes and the Cancer Gene Census (CGC, https://cancer.sanger.ac.uk/census/ ) gene list downloaded on Jan 1, 2023. Therefore, the accuracy of DriverGene is defined as the proportion of genes in the CGC list to all identified genes as follows where Acc i is the accuracy of cancer i , C i is the set of genes identified by DriverGene in cancer i and also in CGC, A i is the set of all genes that identified by DriverGene in cancer i , ∣ ⋅ ∣ is the cardinality of gene set.…”

Trans-Driver: a deep learning approach for cancer driver gene discovery with multi-omics data

“…Four drivers were successfully identified on CHOL, two on KICH, and five on MESO. We also performed an enrichment analysis of two novel methods published in 2020 [44, 45]across 33 cancer types (see Table S3 for details), and the performance of Trans-Driver is still better than the two novel methods.…”

Revisiting the identification of breast cancer tumour suppressor genes defined by copy number loss of the long arm of chromosome 16

“…However there is a class of TSG where reduction of gene dosage from two copies to a single copy is sufficient to reduce transcriptional activity and compromise the normal functioning of the gene in tumours (24). There are no highly mutated genes in breast cancer located on 16q (25) although BC driver genes, CBFB, CDH1, CTCF and NUP93 , have been identified (, 26). It is suggested that loss of transcriptional activity associated with copy number loss is likely to be the basis for functionally compromising the function of TSGs in this region.…”