BERT4DTI : BERT-based Model for Predicting Drug-Protein Interactions

2026-09-27

AI Summary

This study addresses the challenges of scarce labeled data, high fine-tuning costs, and the lack of interactive dependencies in independent sequence encoding for drug-target prediction by proposing an efficient predictive framework. The method employs ChemBERTa and ProtBERT as encoders and introduces a bidirectional cross-attention mechanism to capture specific dependencies between drug-protein pairs, followed by convolutional layers and a multilayer perceptron for classification. Additionally, a truncated layer strategy is adopted to substantially reduce the number of trainable parameters. Experimental results demonstrate that the proposed model achieves state-of-the-art ROC-AUC and sensitivity across multiple benchmarks while reducing the total parameter count to 125 million, thereby realizing an excellent balance between predictive performance and computational efficiency.

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.

Problem

• Drug-Target Interaction
• Sequence-based prediction
• Data scarcity
• Fine-tuning cost
• Pair-specific dependencies

Innovation

• Drug-Protein Interaction
• Bidirectional Mutual Attention
• Parameter-Efficient Fine-Tuning
• ChemBERTa
• ProtBERT