Phenotype to function: using quantitative behavioural fingerprints to predict compound mode of action
File(s)
Author(s)
McDermott-Rouse, Adam
Type
Thesis
Abstract
Phenotypic screens are used to discover novel lead compounds for use as pesticides and anthelmintics as well as pharmaceuticals in some cases. However, these screens rarely elucidate the drug target directly which can make it difficult to understand a compound’s specificity, possible off-target effects including toxicity, and the likelihood of resistance in target species. Finding and identifying these target modes of action requires substantial effort across several disciplines including direct observation of drug effects on animals, sometimes called manual symptomology. As manual symptomology is lengthy and operator dependant, it is a point at which the process could be improved. By using high-resolution high-throughput imaging rigs to collect detailed phenotypic responses from C. elegans treated with test compounds, we detect subtle phenotypes not easily identified by eye. We hypothesised that collecting behavioural phenotypes from drugs where the target is known will allow a comparative analysis with the behavioural effects of unseen compounds and make it possible to predict their modes of action. Here we tested a library of 110 insecticides and anthelmintics on C. elegans, from a panel of 12 wild isolates. We observed strong phenotypes that were shared within mode of action classes particularly for vesicular acetylcholine transporters presenting a coiling phenotype. Furthermore, we trained a classifier which predicts the mode of action to an accuracy of 88% for a subset of 10 modes of actions for the blinded section of the library. We also tested a smaller set of compounds on the plant parasitic nematode H. schachtii and found quantitative differences in their responses compared to C. elegans. Using this approach in tandem with traditional drug discovery routes may lead to faster target discovery.
Version
Open Access
Date Issued
2022-11
Date Awarded
2024-01
Copyright Statement
Creative Commons Attribution Licence
License URL
Advisor
Brown, Andre
Publisher Department
Institute of Clinical Sciences
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)