Mihir Hurwanth
PhD Researcher at Queen Mary University of London
Plants may be among our richest sources of useful chemistry, yet most of it is still uncharted — we have recorded compounds from only a small fraction of species, and rarely at random. My PhD asks how large that gap really is, how biased our current picture of plant chemistry is, and whether a species' ecology can tell us where to look next.
To study this, I build open data pipelines that harmonise the major natural-product databases into a single, structure-keyed catalogue of plant chemistry, linked to Kew's taxonomy, global distributions and the plant tree of life. With it I measure how non-randomly plants have been sampled — by human use, taxonomy, evolutionary history, geography and conservation status — which suggests that widely cited totals for plant chemical diversity are better read as bias-conditioned lower bounds than as complete counts. I then use species-richness estimators to put a range on how much may still be missing.
A second strand tests a simple idea: that choosing which species to screen by their ecological diversity — their traits, environments and evolutionary distance — may accumulate new compounds faster than screening at random. I am now extending this into a multimodal model that brings together herbarium images, phylogeny and trait–environment data to predict the medicinal potential of the many plant species never chemically assessed.
What motivates me is connecting two worlds that rarely meet — global biodiversity patterns and the fine-grained logic of drug discovery. By learning from how plants have solved chemical problems over millions of years, I hope to make early-stage discovery more predictive, more interpretable, and better aligned with the biology and ecology that underpin it.