Mutations can affect many behavioural traits and, although improving one, might worsen many others. One hypothesis suggests that this is a particular issue in animal behaviour if based on many degrees of freedom: diverse environmental effects can lead to different phenotypes for the same genotype, i.e. many-to-many genotype-phenotype mapping. An alternative hypothesis suggests that the phenotypic behavioural space is rather low dimensional and hence tolerant to genetic changes: different genotypes can give rise to the same phenotype, i.e. many-to-few genotype-phenotype mapping – a result of neutral mutations, which are neither beneficial nor detrimental to the organism’s fitness. Until now, these conflicting hypotheses have not been tested rigorously in species other than bacteria, viruses and plants. Moreover, they have never been tested in a social context due to challenges of connecting changes in genes with specific social behavioural phenotypes . This was mainly because the animals whose collective behaviour has received the most attention, such as swarming ants, schooling fish, and flocking birds, are not particularly amenable to genetic manipulation. Furthermore, extracting a behavioural phenotype requires an accurate representation of the fine-scale dynamics of individual motion and interactions, not always approachable even with modern tracking techniques. Judging from these points, the nematode worm Caenorhabditis elegans is a perfect system for genotype-phenotype mapping: it displays rich social dynamics during feeding (so-called collective feeding), it is genetically tractable, and owing to recent advances in robotic imaging, completely trackable, even in large groups. For the first time, all these advantages will be combined in a single project, allowing to test the hypotheses mentioned above in a rigorous and structured way using quantitative phenotyping and computational modelling.
The aim of this project is to quantify the genotype-phenotype mapping in C. elegans social behaviour using quantitative collective modelling based on high-throughput tracking data. The specific goals describe a three-stage procedure:
Goal 1: Develop a dictionary of worm behavioural states. This goal will be accomplished in two steps: I will first identify various behavioural states based on worm postures and tracking data and then quantify transitions between the different worm states within this behavioural map.
Goal 2: Build an agent-based model (ABM) and refine it to match the experimental summary statistics. This work will be also done in two steps: I will build an ABM that captures multiple behavioural worm states and subsequently parameterise and refine it to match the experimental summary statistics.
Goal 3: Derive and assess genotype-phenotype mapping by extracting a low-dimensional representation of the model. This is the most ambitious and risky goal. To achieve it, I will extract phenotypes and test their similarity among almost 200 C. elegans strains.
