Cancer develops through the acquisition of multiple mutations, and it is assumed that genetic interactions between mutated genes play an important role in cancer onset and progression. One approach to find genetic interactions in cancer is to search for pairs of mutated genes in tumors that occur together more (or less) often than expected given the frequency of the individual mutated genes. Highly co-occurring mutated genes suggest a cooperative role of these altered genes in cancer development. Mutually exclusive gene pairs are possible examples of synthetic lethality, and are therefore interesting for the development of new cancer treatments.
We are working on the development of a computational pipeline to detect such significant cases of co-occurrence and mutual exclusivity in large pediatric cancer data sets. We focus our search on candidate gene pairs in specific cancer types as well as those pairs that are only found when all cancer types are combined in one set. Particularly challenging in our analysis is to distinguish between real candidates and false positives that can arise due to technical artifacts or incorrect assumptions about the background expectation.
In a follow-up phase, we will validate our top candidates experimentally to confirm the co-operative nature of co-occurring genes and test the cell-fate of double knock-outs of mutually exclusive genes.