Karen's team wins Cancer Research UK "AI in Cancer Omics" Hackathon

Our MPhil student Karen Lee participated in the CRUK "AI in Cancer Omics" hackathon from the 9th to 11th of July. This year, the hackathon focused on how to leverage large amounts of data to learn about cancer resistance mechanisms from species (e.g. bowhead whales, elephants) that exhibit remarkable cancer resistance despite their longevity and body size. Karen's team approached the problem from the angle of convergent evolution, by asking: If cancer resistance evolved independently across species, do they share common mechanisms, and can these be detected through convergent amino acid evolution?

Karen said:

Our project would not have been possible without Evolutionary Sparse Learning with Paired Species Contrast (ESL-PSC), a method developed by John Allard, Sudhir Kumar and others at Temple University, who also served as incredible mentors to our team during the hackathon. ESL-PSC is a genetic model that identifies genes and amino acid sites with convergent molecular evolution associated with convergent trait acquisition (in our case, cancer resistance). Applying their method, we identified several genes and amino acid sites that predicted cancer resistance in independent held-out species. In an extension, we explored replacing amino acid multiple sequence alignment inputs with 3Di structural sequence alignments derived from ESMFold predictions, finding that linearized structural representations of proteins result in promising prediction accuracy as well as new information on targetable protein sites. It was really enriching to collaborate with people from different research backgrounds, and I learned a lot both from my team and others. Thanks so much to my teammates, Anna-Maria Papameletiou, Tomás Lopes, Samuel Innes, Riham Hlis and Peter Oluwatimilehin and our mentor John Allard for a energizing weekend of collaboration. Our combined efforts culminated to a first place win! Thanks so much as well to the hackathon organizers Eloïse Trabut, Alex Cagan and all others involved in making this event possible.

Many congratulations to Karen and her team!