Statistical learning in molecular medicine
We develop methods to integrate multi-omics and other data, e.g. clinical data or drug characteristics, for predicting a patient's prognosis or treatment response. In this way we contribute to the development of novel cancer treatments.
Estimation of drug efficacy and synergistic effects in a drug combination experiment. The plots were generated with the R package bayesynergy v2.1 for an example data set from O'Neil et.al, 2016, DOI:10.1158/1535-7163.MCT-15-0843.
About the group
We develop new multivariate (multi-task) penalized and Bayesian methods that allow us to integrate multi-omics data and other data to predict a patient's prognosis or treatment response with a focus on applications in personalized cancer therapy.
In many of our projects we develop methods for estimation and prediction of drug sensitivity and drug synergy in large-scale in vitro pharmacogenomic screens, based on molecular characterization of cancer cell lines and patient samples as well as properties of the drugs. The aim is to guide the selection of cancer therapy by statistical prediction of how drugs will behave for the individual patient, both for each drug on its own and for the combination.
Projects
- Multivariate structured penalized and Bayesian regressions for pharmacogenomic screens
-
Bayesian semi-parametric models to estimate synergistic interaction effects in in-vitro drug combination experiments
-
Prediction of cancer drug combinations in large-scale pharmacogenomic screens
Cooperation
We cooperate with both national and international researchers:
- Several collaborations with colleagues at OCBE (UiO) and at Oslo University Hospital
- International collaborations on modelling and prediction of drug sensitivity and synergistic effects in pharmacogenomic screens: Paul Kirk and Sylvia Richardson (MRC Biostatistics, Cambridge, UK)
- International collaborations on statistical integration of multi-omics and other data sources for risk prediction in cancer: Alex Lewin (Dept of Medical Statistics, London School of Hygiene and Tropical Medicine, UK), Axel Benner (Dept of Biostatistics, German Cancer Research Center), Katja Ickstadt and Jörg Rahnenführer (Faculty of Statistics, TU Dortmund University, Germany)