Areas of expertise:
C. elegans; Behavioural Phenomics; High-throughput Phenotyping; Behavioural Tracking; Drug Screening; Rare Genetic Diseases; Neurological Diseases; Epilepsy; Protein Engineering; Structural Biology
I am cross-disciplinary biologist who sits at the interface of fundamental research and its translation into the clinic thanks to my uniquely specialised combination of wet-lab and computational skills. My interests are focused on developing high-throughput computer vision-led approaches for tracking the behaviour (phenotyping) of the nematode round roundworm, Caenorhabditis elegans, as a proxy for studying and developing treatments against human genetic diseases, particularly epilepsy and other neuromuscular disorders. I completed an integrated Bsc and Masters in microbiology at the University of Leeds (2017) before obtaining a PhD in biochemistry from the University of Cambridge (2021). I then spent 3 years in a postdoctoral position at Imperial College London and the MRC Laboratory of Medical Sciences, as member of Andre Brown’s group, before returning to Leeds as a Wellcome Trust funded Early Career Research Fellow.
I am constantly looking to form new collaborative links with clinicians, industrial partners and disease foundations. To date, I have modelled >100 genetic variants associated with human diseases observed within the clinic. As part of ongoing patient-led collaborations I have conducted numerous large-scale drug repurposing screens (>6000 compounds) for the treatment of mutations in TNPO2 (supported by the TNPO2 foundation), ARID1B (supported by the Foundation for ARID1B Research) and GNAO1 (supported by the Undiagnosed Diseases Network) mutations. I am also a member of a world-wide consortium of yeast, fly and worm researchers (mediated by Perlara PBC) who have successfully identified drug repurposing treatments for GNAO1 mutations that are currently in Phase III clinical trials.
Research interests
My research group are interested in genotype-phenotype mapping the connection between genetic variation (mutations associated with human disease) or chemical perturbation (exposure to, or treatment with, bioactive compounds/molecules) and phenotypic variation (changes in behaviour). Primarily, we are focused on using automated high-throughput deep-phenotyping of C. elegans to develop an in vivo screening platform that can be universally applied for studying, and identifying candidate therapeutics for the treatment of any genetic disease, provided a conserved C. elegans ortholog exists.
Disease modelling as genotype-phenotype mapping in humans and C. elegans. Arrows show the progression from: (1) symptom identification (a human phenotype outside the healthy range, i.e. disease) to (2) genotyping (placing a patient in genotype space by identifying a genetic variant), (3) disease model creation (making a corresponding mutation in a model organism), and (4) model organism phenotyping. Phenotypic drug screens identify compounds that move a disease model toward the wild-type phenotype and constitute lead therapeutic compounds (green arrows). Read more about my approach here.
Historically, drugs were serendipitously discovered by observing their effects on an organism, typically humans or animals, and this is particularly true for compounds which exert an effect on behaviour. In the era of targeted drug development, phenotypic screening continues to play an important role in drug discovery efforts. Such screens often rely on a strong behavioural change as the result of genetic mutation or exposure to a compound. Yet, the majority of neurological conditions/treatment with bioactive compounds do not give rise to readily-observed phenotypic alteration(s). Thus, conducting large-scale drug screens is expensive, laborious and timely, making the development of bespoke treatments for a specific genetic variant near impossible. This is particularly true for rare (often underfunded) disorders, where a validated target for the development of candidate therapeutics rarely exists.
This gives rise to the question: “How can we make phenotypic screens more effective?”
My approach combines the use of a novel high-throughput C. elegans imaging system, capable of recording the behaviour of many worms in parallel, with automated quantitative behavioural phenotyping methods. Using a computational ethology (computer vision combined with machine learning) approach, we can accurately quantify multi-dimensional changes in the locomotion, morphology and posture of C. elegans following genetic mutation/chemical perturbation that would otherwise be missed via observation with the naked eye.
Using genome editing, we develop C. elegans strains that contain mutations orthologous to those associated with diseases observed within the clinic. These disease model mutants can be thought of as “patient avatars” of disease. By comparing the behavioural state of a patient avatar to “healthy” wild-type (the parental strain used to make the patient avatar), we now have a readily-screenable method for directly quantifying the sickness of a mutant, and for assessing how treatment with a drug/biologic can improve/worsen the disease state of the patient avatar strain (see the figure below).
This approach is beneficial for several reasons:
Speed: any standard sized multiwell plate can be assayed in just 15 minutes
Universal applicability: utilising a simple standardised assay for phenotyping any mutation or treatment condition means no prior knowledge of how a genetic variant causes disease, or the mechanism of drug action, is required to discern a behavioural phenotype and/or conduct a high-throughput drug screen [my approach can be readily applied in the study of ultra rare diseases and unknown genetic variants]
Cost: the generation and screening of transgenic C. elegans strains is much cheaper (and often faster) than using other whole organism models, e.g., mice
Multidimensionality: extracting high-content behavioural “fingerprints” of phenotypic differences (>8000 behavioural features per strain/treatment condition) allows: (1) the accurate identification of subtle behavioural changes, (2) permits an assessment of off-target/side effects of drug treatment during an initial compound screen [removal of nuisance compounds early in a drug screen greatly enhances the translation of hits towards the clinic]
In the above figure, the “healthy” wild-type strain, N2, is denoted by the blue star. The “sick” mutant, unc-80 (containing loss-of-function of the worm UNC80 protein, akin to UNC80 defeciency in humans), is denoted by the red star. Each dot represents the patient avatar worm strain (unc-80 loss-of-function mutant) treated with compounds from an FDA-approved drug library. This demonstrates a movement towards and away from the healthy strain in phenotypic space upon drug treatment. The entire library of ~750 drugs was screened in ~1 hour, and the blue dots represent the top hits that can be taken forward for confirmation screening.
You can find an automatically updated list of my publications here.
<h4>Research projects</h4>
<p>Some research projects I'm currently working on, or have worked on, will be listed below. Our list of all <a href="https://biologicalsciences.leeds.ac.uk/dir/research-projects">research projects</a> allows you to view and search the full list of projects in the faculty.</p>
