Biomedical AI

We are interested in data-driven precision medicine. With the advent of screening technologies like DNA or RNA sequencing, metabolomics, etc, we can now get comprehensive maps of the molecular composition of the cells in our body. Our general aim in this theme is to develop methods to extract information from such data and use it to improve therapies. Diego wrote a short piece on the topic for the World Economic Forum.

Our main research activities in this space are:

• Machine learning for drug discovery. We are using supervised learning to discover new drugs for cancer therapy, in collaboration with JC Acosta’s lab and Neil Carragher’s team.

• Genotype-phenotype mapping. Using a range of learning methods, we aim at characterising sequence-function relations in several systems, including RNA folding (with Greg Kudla’s lab) and protein expression (with Guillaume Cambray’s lab).

• Time series analysis. We use a mix of regression methods and graph-based clustering to identify temporal gene expression patterns. For example, with Holger Auner’s lab we studied the transcriptomic response of cancer cells to protease inhibitors.

• Cancer metabolism. We are developing network-based methods to detect metabolic drug targets, with a mix of network theory and genome-scale modelling aimed at detecting metabolic nodes that could be used for therapy or diagnostics. This project is in collaborations with the Cancer Metabolomics group and the Centre for Mathematics of Precision Healthcare at Imperial College London.

We also work with policy makers and provide science-based advice for the implementation of precision medicine globally. Diego sits on the Scientific Advisory Board for Precision Medicine of the World Economic Forum Centre for the Fourth Industrial Revolution, based in San Francisco.