Historically, new drugs were discovered by directly observing their effects in humans or animals, often serendipitously. This is especially the case for drugs with effects on behaviour. See this fascinating history of psychopharmacology for some examples. Even in the era of targeted drug discovery, phenotypic screening has continued to play a prominent role in drug discovery. So the question is, how do we make phenotypic screening even better? Our approach has been to build new imaging systems with arrays of high resolution cameras so that we can record many worms in parallel in multiwell plates. We then take a computational ethology approach to identify the best ways of detecting the symptoms of mutant worms and the effects of drugs or natural products on behaviour.

Finding the Right Molecules

An important part of this approach is finding the right source of molecules. Drugs that are already approved for human use are attractive because they are already known to be safe in humans. Therefore, if one of them is found to be useful in treating a disease-model worm, there is a better chance that it will be translatable to human use than compounds from some other sources. At the same time, we are also very excited about the prospects of phenotypic screening for discovering neuroactive molecules from nature that are difficult to find using other methods. We are therefore looking at natural product libraries, soil and gut microbiomes, and diverse venoms.

Target Identification

A perenial difficulty with phenotypic screens is target identification. That is, given a hit in a phenotypic screen, how do you figure out what targets the compound is affecting? We are using machine learning approaches and a database of worm drug phenotypes to predict targets based only on worm behaviour. In parallel, we will also use worm genetics to identify targets, which has proven to be a successful approach in many previous studies.