innovating high-throughout toxicity testing.

For my Neuroscience honours thesis at the University of Edinburgh, I designed and executed a high-throughput in vivo screen to evaluate 38 drug candidates for progressive multiple sclerosis (PMS) — a stage of the disease marked by neurodegeneration and currently lacking effective treatments.

Using larval zebrafish, I assessed drug-induced toxicity across 18 physiological and morphological endpoints, generating over 23,000 datapoints. To overcome the interpretive bottlenecks of multi-endpoint screens, I developed a novel, modular toxicity index that captures cumulative, multidimensional toxicity as a single summary score — something existing indices (e.g. LD50 or LOADED) fail to do.

I wrote custom Python tools to automate both toxicity quantification and index calculation

To further optimize the statistical evaluation of high-content drug data, I implemented Bayesian hierarchical linear models — rarely used in preclinical zebrafish studies — to extract compound-level effects while reducing false positives and accommodating uncertainty. I also used PCA to identify mechanistic toxicity clusters and reveal that surface area changes could act as a proxy for classifying toxicity type (developmental vs general).

Finally, in follow-up efficacy assays targeting microglia and oligodendrocytes, I identified a potential synergistic effect between Guanabenz and Pomalidomide, which increased oligodendrocyte numbers only when combined — supporting the case for rational combination therapies in neurodegenerative disease.