BERT based model improves prediction of drug protein interactions
BERT4DTI : BERT-based Model for Predicting Drug-Protein Interactions
Machine LearningArtificial IntelligenceComputation and Language
Summary
Predicting if a drug will interact with a specific protein is important for finding new medicines and repurposing existing ones. The authors introduced BERT4DTI, a model that uses advanced language techniques to better understand the chemical and protein sequences together, instead of separately. This approach helps the model make more accurate predictions while being faster to train than fully fine-tuning big pretrained models. The model showed strong results on several standard tests, especially in identifying true interactions.
What this means in practice
- •For pharmaceutical data scientists: Provide an efficient model to screen potential drug-protein pairs for interaction using sequence data with fewer training resources.
- •For bioinformatics engineers: Integrate a sequence-based interaction prediction method that captures joint drug-protein dependencies for early-stage drug discovery pipelines.
Authors
Thanina Hamitouch, Khadidja Henni, Abdelkrim Arie, Amina Selma Haichour, Neila Mezghani, Lina Abou-Abbas
Abstract
Understanding how drugs interact with protein targets is fundamental to drug discovery, drug repurposing and the early identification of promising therapeutic candidates before costly experimental testing. Sequence-based DTI models face three practical limitations: labelled interactions are scarce and unevenly distributed, large pretrained chemical and protein encoders are expensive to fine-tune end-to-end, and independently encoded sequences do not capture pair-specific dependencies. We present BERT4DTI, which encodes SMILES strings with ChemBERTa and amino-acid sequences with ProtBERT, applies bidirectional mutual attention between token-level representations, and classifies the resulting interaction features using convolutional layers and a multilayer perceptron. To reduce trainable size, ProtBERT is truncated to 18 retained layers and only the last two layers of each encoder are fine-tuned. On BIOSNAP, DAVIS and BindingDB, BERT4DTI is competitive, achieving the best ROC-AUC and PR-AUC on BIOSNAP and the highest sensitivity on all three benchmarks. An ablation on DAVIS shows that mutual attention improves PR-AUC and specificity. With 125M trainable parameters compared with 353M for full BERT fine-tuning, BERT4DTI provides a favourable performance-parameter trade-off for sequence-based DTI screening, while leaving runtime profiling, calibration and leakage-audited validation for future work.