A multi-modal co-attention model for accurate drug-target interaction prediction
Wanjun Ma, Wenjun Li, Mengyun Yang, Zhengdong Pu, Xiwei Tang · 2026
Read A multi-modal co-attention model for accurate drug-target interaction prediction by Wanjun Ma, Wenjun Li, Mengyun Yang, Zhengdong Pu, Xiwei Tang (2026) 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/plos-pone-0351880.txt.
Description
Wanjun Ma and colleagues built a neural network that predicts whether a drug will latch onto a given protein, a first question in developing and repurposing medicines. Older models read the drug's structure or the protein's sequence and guess from one side. Their model reads both at once, with a co-attention step that lets each side highlight the parts of the other it should weigh.
On three standard benchmarks the model beat leading rivals on the usual accuracy scores, reaching very high marks on the easiest set. Removing the co-attention step hurt performance, showing the fusion does real work. High-confidence predictions the team inspected also lined up with known biology. For drug hunters, a sharper first filter means fewer dead ends carried into expensive laboratory testing.
Catalog record
| CREATOR | Wanjun Ma, Wenjun Li, Mengyun Yang, Zhengdong Pu, Xiwei Tang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 46190 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |