DR-066 Googling the cancer genome

Cancer affects millions of people worldwide. Current DNA sequencing technologies allow routine screening of patients for cancer-associated mutations that can be targeted in personalized cancer treatments. Accurate computational methods for the identification of mutations, in particular structural variants (SVs), are lacking, together with methods that can prioritize mutations for further intervention. To solve this problem, we are developing a new set of machine learning approaches that can learn to detect SVs directly from ‘raw’ sequencing data and prioritize them for clinical intervention based on multiple sources of genomic information. Our work will open up a new avenue of machine-learned SV detection and prioritization, and accelerate the introduction of SV-based screenings for new treatments in clinical settings.

Jeroen de Ridder UMC Utrecht the Netherlands