(1/5) The latest from our group: Jenna Fromer's overview of computer-aided multi-objective optimization in small molecule discovery is now online & open access @Patterns_CP | doi.org/10.1016/j.patt… #compchem (2/5) Our focus here is on Pareto optimization. Pareto optimization introduces additional algorithmic complexities, but reveals more information about the trade-offs between objectives and is more robust than scalarization approaches

1/ Learning compound-protein interactions (CPI) w/ sequence & compound alone is tantalizing, but time and time again, CPI models fail to beat simple baselines. Does this paper do so successfully? An analysis from @samgoldman19 :

https://twitter.com/KevinKaichuang/status/1543942451848790018

2/ The short answer is no, this model also fails to outperform a simple nearest neighbor baseline. The metabolic models presented in the paper are likely independently valuable, but are *not* enabled by deep learning