Deep Learning the Terpenome: Unpacking Plant Chemical Arsenals

For thousands of years plant-derived natural products have been harvested for their medicinal properties in an effort that is now called Bioprospecting. The astonishing chemical diversity of biologically active substances in various organisms reflects an equally staggering diversity infunction. Despite enormous diversity, most phytochemicals are terpenes, derived from five-carbon isoprene units assembled and modified in thousands of ways, such that the full complement of genes involved in terpenebiosynthesis is now called the ‘Terpenome’. Understanding the mechanism by which few initial substrates are converted into a tremendous chemical arsenal holds promise for improving agro-biochemical and pharmacological potential of plants.  

My talk will be about how we push and merge boundaries of modern and traditional technologies like deep learning and headspace chromatography, to investigate this plant chemical production line. We use supervised and unsupervised machine learning to correlate the plant chemotype with its phenotype and genotype, we also use complex network analyses to enable transcriptional regulatory inferences from large scale gene expression datasets. At the end of my talk, I hope to emphasize the role of University Herbaria and their specimen collections from various parts of the world spanning centuries, and how these offer untapped opportunities to expand botanical investigations by covering very wide taxonomic and spatio-temporal scales. 

The seminar will be hosted using Zoom. A Google calendar invite featuring the Zoom link will be sent to Biology staff and students before the seminar date. For all enquiries please contact Biology DMT Hub.