A big data approach to cancer driver genes yields insights into how these mutations affect protein interaction
A collaborative effort between researchers in Europe and the US has resulted in the combination of two publicly available “omics” databases, to create a catalog of cancer drivers. The study has discovered over 70 new candidate cancer driver genes (1), and could help to explain how the same affected gene can lead to different outcomes or therapy responses in patients. The computational program, called E-driver, uses tumor data from ~6,000 patients from the Cancer Genome Atlas (TCGA), and more than 18,000 three-dimensional protein structures from the Protein Data Bank. An algorithm then analyzes the information to see if structural alterations of protein-protein interaction (PPI) interfaces are enriched in cancer mutations, therefore identifying candidate driver genes.
The motivation behind the study? It was based on the theory that, as genes can have a variety of functions, information on the structures, pathways, and protein complexes involved in disease would give insight into how mutations in different genetic regions may produce different characteristics in the resulting cancer. “I was surprised that almost all existing tools for the analysis of cancer mutations are not using available structural information on proteins, which happen to be my main field of interest,” says co-author Adam Godzik. “Insights can be gained from even very rudimentary structural analysis, and we set out to do this on a large scale”.
As well as identifying possible new cancer drivers, the study has given further insight into this area of oncology, adds Godzik. “We have learned two things: mutations in different regions of a gene can have different, sometimes opposite effects on cancer and treatment outcomes. And the growing list of cancer driver genes is slowly eroding the current model of driver vs. passenger mutations. It is clear now that as well as a small number of major drivers, there are a lot of genes that play a role of ‘enablers’ or ‘mini-drivers’, which when mutated, could provide an important advantage to a growing tumor, but may not be able to start cancer by themselves”, he says. Godzik admits that more analysis is needed to better understand the entire landscape of molecular events in cancer in order to identify optimal treatments and predict patient outcomes. However, at this point it remains unclear if these newly-discovered drivers are likely to become targets for therapy, as many are relatively rare.
Newsletters
Receive the latest pathology news, personalities, education, and career development – weekly to your inbox.
References
- E Porta-Pardo et al., “A pan-cancer catalogue of cancer driver protein interaction interfaces”, PloS Comput Biol, 11, e1004518 (2015). PMID: 26485003.