Docking-Aware Attention: Dynamic Protein Representations through Molecular Context Integration
Amitay Sicherman, Kira Radinsky · 2025
Read Docking-Aware Attention: Dynamic Protein Representations through Molecular Context Integration by Amitay Sicherman, Kira Radinsky (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/arxiv-2502-01461.txt.
Description
Amitay Sicherman and Kira Radinsky build a protein model that changes its mind depending on the molecule it faces. Their 2025 paper tackles reaction prediction for enzymes, the proteins that drive chemical change in drug design and green chemistry. One enzyme often handles many partners, yet older models either skip protein structure or freeze the protein into one fixed summary.
Their architecture, Docking-Aware Attention, blends docking scores, which estimate how a partner sits on the protein, with learned attention weights. The blend lets the model spotlight the protein regions that matter for each partner. On enzyme reaction tests it reached 62.2 percent accuracy on complex molecules, up from 56.79 percent for the prior best. Gains also held on novel reactions. The attention maps can be read back to see why a call was made.
Catalog record
| CREATOR | Amitay Sicherman, Kira Radinsky |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-01461