Each 'bird's nest' in the image a single worm's trajectory through the shape space defined by three eigenworms (colour indicates position in the fourth dimension).


The goal of behavioural genomics is to understand the mapping between genome variation and behaviour. Despite rapid advances in technology for sequencing and engineering genomes, it is still a challenge to associate particular genes with heritable behavioural differences because behaviour is time consuming to measure and difficult to quantify. To address this, we are developing high-throughput imaging platforms to capture the complex behaviour of thousands of sequenced strains of the nematode C. elegans.

Higher throughput and resolution

We are always looking for ways to improve animal tracking. One way to do this is to build better imaging systems that can record from more individuals at higher spatial and temporal resolutions. We can piggyback on developments in sensor technology driven by the mobile phone industry and use high-resolution cameras to record many worms at the same time without sacrificing out ability to resolve subtle postural changes. Right now we're building a rig with multiple cameras that will combine the advantages of high-res multi-worm tracking (plus some improvements!) with those of running many cameras in parallel. We've also included some automation to reduce operator labour.

Better quantitative phenotyping through physics

Higher throughput video recording will produce even larger datasets than those we've collected in the past and so another major aim of the group is to process video of behaviour in an unbiased way and to understand what it all means. Scientifically, we're interested in questions about the complexity of behaviour, how behaviour is organised by the nervous system, and how it evolves.

We think about dynamical systems, finding patterns using data compression, and exploring the analogy between language and behaviour using methods from natural language processing.

Making the connection to genetics

A better quantitative understanding of behaviour will help us develop more sensitive and informative phenotypes. Increasing sensitivity means we will be able to detect more subtle differences between mutants and increasing informativeness means we will have a better idea where to focus our functional experiments when we're following up on hits from screens.

Recombinant inbred lines are a great way of looking at the genetics of complex traits with some notable recent successes in worms. We're using automated imaging to perform quantitative trait locus mapping at a larger scale than has been possible until now. Having high-dimensional quantitative phenotypes will maximise the mapping resolution and hopefully lead to the identification of many new genes that affect behaviour.

Complex behaviour in worms

Worms are usually thought of as simple organisms, and indeed their nervous systems are incredibly small. Still, worms display some complex behaviours. For example, there are strains that aggregate into groups on food and others that remain solitary. This difference in collective behaviour is known to be under genetic and environmental control, but the behaviour itself has not been studied sufficiently. A biophysical approach will give us a better understanding of the behaviour itself and a deeper look at the genetic and neural circuit mechanisms that govern animal interactions.