Phenotype to Function: Predicting compound mode of action from behavioural fingerprints

Abstract

Pesticides and anthelmintics (nematode-killing drugs) are discovered through phenotypic screens in target species.  This means that their efficacy is often known early in the development pipeline, but their mode of action is not.  Therefore, an important problem in developing new compounds to combat the rise of anthelmintic resistance is determining their mode of action.  We are using a combination of highthroughput imaging, quantitative phenotyping, and machine learning to try to predict a compound's mode of action from its effects on C. elegans behaviour.  I will present our approach, recent results, and some conceptual challenges in dealing with multi-dose data sets.

Bio

My education is in physics (BSc. Memorial University of Newfoundland, PhD University of Pennsylvania) but I've always been fascinated by biology.  I started working with C. elegans during my postdoc at the Lab of Molecular Biology in Cambridge with the idea that some of the approaches that had been useful in molecular and cellular biophysics might also be useful at the whole-animal level.  Since 2013 I've been a group leader at the MRC London Institute of Medical Sciences at Imperial College London.  My lab works on the physics of behaviour with a particular focus on how better representations of behaviour can help us understand the genotype-phenotype map and how neuroactive compounds affect behaviour.