Can AI-Predicted Complexes Teach Machine Learning to Compute Drug Binding Affinity?

Wei-Tse Hsu, Savva Grevtsev, Anna M. Herz, Thomas Douglas, Aniket Magarkar, Philip C. Biggin · 2025

Read Can AI-Predicted Complexes Teach Machine Learning to Compute Drug Binding Affinity? by Wei-Tse Hsu, Savva Grevtsev, Anna M. Herz, Thomas Douglas, Aniket Magarkar, Philip C. Biggin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-12728920.txt.

Description

Wei-Tse Hsu and co-authors ask whether AI predicted protein drug complexes can train scoring models that estimate binding strength, in work from twenty twenty-five. Machine learning scorers need many example complexes, and measured structures are scarce. Co-folding models can mint synthetic ones, but quality varies.

The team found that gains from synthetic training data hinge on the structural quality of the added complexes. Poor predictions drag the scorer down rather than lift it. They set simple heuristics that flag high quality predictions without any reference structure to check against. Those filtered predictions can stand in for experimental structures in training. The study guides future augmentation plans that lean on co-folding.

Catalog record

CREATORWei-Tse Hsu, Savva Grevtsev, Anna M. Herz, Thomas Douglas, Aniket Magarkar, Philip C. Biggin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE20 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC12728920

Browse the full library · About the Tower of Babel