Michael Brocidiacono
@brocidiacono.bsky.social
I teach chemistry to computers
Check it out here: arxiv.org/abs/2506.00593
Look mom, no experimental data! Learning to score protein-ligand interactions from simulations
Despite recent advances in protein-ligand structure prediction, deep learning methods remain limited in their ability to accurately predict binding affinities, particularly for novel protein targets d...
arxiv.org
June 4, 2025 at 7:01 PM
Check it out here: arxiv.org/abs/2506.00593
Using this workflow, we're getting competitive performance at virtual screening across 6 targets -- but this is just the beginning. This is the first AI method that has the potential to work on even the weirdest target classes, such as disordered proteins or even RNA! (3/)
June 4, 2025 at 7:01 PM
Using this workflow, we're getting competitive performance at virtual screening across 6 targets -- but this is just the beginning. This is the first AI method that has the potential to work on even the weirdest target classes, such as disordered proteins or even RNA! (3/)
What if we can generate the data for a new target? We show that you can use molecular dynamics simulations of your target in complex with random small molecules to train a target-specific model via force-matching. (2/)
June 4, 2025 at 7:01 PM
What if we can generate the data for a new target? We show that you can use molecular dynamics simulations of your target in complex with random small molecules to train a target-specific model via force-matching. (2/)