TransformerCPI: improving compound–protein interaction prediction by sequence-based deep learning with self-attention mechanism and label reversal experiments
importance 5/10
partially solved
Computational Drug Discovery Methods drug discovery stage: research open-source model
Problem
Predicting compound-protein interactions from protein sequence alone, while avoiding dataset pitfalls and hidden ligand bias that overestimate performance.
Why it matters
CPI prediction guides drug discovery, and many target proteins lack 3D structure data.
What the AI did
Built TransformerCPI, a transformer neural network with self-attention, trained on newly constructed datasets and validated with a stricter label reversal experiment.
Results
Much improved performance on the new experiments (no numbers given in abstract)
Impact
Provided an improved sequence-based CPI predictor and more rigorous evaluation practices.
Record
| Venue | Bioinformatics |
|---|---|
| Published | 2020 |
| Work type | article |
| Authors | Lifan Chen, Xiaoqin Tan, Dingyan Wang, Feisheng Zhong, Xiaohong Liu, Tianbiao Yang, Xiaomin Luo, Kaixian Chen et al. |
| Citations | 597 (OpenAlex, at import time) |
| References | 41 |
| Model or method | TransformerCPI (transformer neural network) |
| Organization | not stated in source |
| Primary source | https://doi.org/10.1093/bioinformatics/btaa524 |
| Open access | bronze |
| Keywords | compound-protein interaction, transformer, self-attention, deep learning, drug discovery, label reversal |
Importance rating
5
real but narrower advance
CPI prediction aids drug discovery; abstract calls it crucial but gives no quantified impact or scale of affected targets.