%0 Generic
%A YANG, MI
%C Heidelberg, Germany
%D 2019
%F heidok:25605
%K bioinformatics, computational biology, machine learning
%R 10.11588/heidok.00025605
%T Applying machine learning to derive actionable insights in precision oncology
%U https://archiv.ub.uni-heidelberg.de/volltextserver/25605/
%X Cancer drugs have among the lowest response rates across all diseases. Combining the wealth of omics data and machine learning is a promising way to reach this goal. In this thesis, we addressed the following aspects of precision oncology: (i) We used Macau, a bayesian multitask multi-relational algorithm to explore the associations between the drugs’ targets and signaling pathways’ activation. We applied this methodology to drug synergy prediction and stratification. (ii) We leveraged through a collaborative machine learning competition to understand the association between genome, transcriptome and proteome in tumors. The main focus of this thesis is to use machine learning to generate actionable insights, for more personalized therapies.