Research Dashboarding: Visualizing cancer mutation patterns

This is was part of a past internship at Genospace in Data Analytics. All patient data is publicly available from the MSK-IMPACT study.

Over the past several months I have been working in an internship role at Genospace, and I had the great opportunity to pursue research while there. And in that research I ended up with an terrific opportunity to employ Shiny analyze and also present my analysis.

Our goal in this research was to seek particular types of mutation patterns in cancer data — specifically mutually exclusive mutations. \Mutually exclusive mutations tend to imply redundant effects, so for instance we don’t see genes “A” and “B” mutate together because both lead to the same effect that gave the patient cancer. Two paths to the same end.

If we see a single gene that is mutually exclusive with a large number of other genes, then we have a smoking gun. This situation implies a single gene with a wide range of cancerous effects — this is what a driver mutation looks like. A single gene causing all deleterious effects of several oncogenes at once. On the plus side, this also means we have excellent target for treatment. Identifying driver mutations like these is a effective way of killing several birds with one stone, and this method is one of several in development.

I was employing one of several statistical tests (called DISCOVER) to identify mutations patterns in cancer, this time looking at data from the Memorial Sloan Kettering’s public MSK-Impact study. However, the results returned by the test were initially fairly difficult and tedious to visualize and to find noticeable patterns. I reasoned that a web application would allow for a far faster and more dynamic method of sifting through the results rather than working exclusively within R code. This also had the added benefit of allowing easier collaboration with colleagues who do not employ R in their work, or for that matter for quick visualizations during presentations.

Take a look at the application below. If you want to do your own analysis on the MSK data, you can find a direct link to the study’s results below as well! Special thanks to Dan Schlauch and others at Genospace for the help and guidance!

Direct Link: https://connorh982.shinyapps.io/driverAnalytics/