Antibody lead discovery, crucial for immunotherapy development, requires identifying candidates with potent binding affinities to target antigens. Recent advances in protein language models have opened promising avenues to tackle this challenge by predicting antibody–antigen interactions (AAIs). Despite their appeals, precisely detecting binding sites (i.e., paratopes and epitopes) within the complex landscape of long-sequence biomolecules remains challenging. Herein, we propose MambaAAI, a bio-inspired model built upon the Mamba architecture, designed to predict AAIs and identify binding sites through selective attention mechanisms. Technically, we employ ESM-2, a pre-trained protein language model to extract evolutionarily enriched representations from input antigen and antibody sequences, which are modeled as residue-level interaction matrixes. Subsequently, a dual-view Mamba encoder is devised to capture important binding patterns, by dynamically learning embeddings of interaction matrixes from both antibody and antigen perspectives. Finally, the learned embeddings are decoded using a multilayer perceptron to output interaction probabilities. MambaAAI provides a unique advantage, relative to prior techniques, in dynamically selecting bio-enhancing residue sites that contribute to AAI prediction. We evaluate MambaAAI on two large-scale antibody–antigen neutralization datasets, and in silico results demonstrate that our method marginally outperforms the state-of-the-art baselines in terms of prediction accuracy, while maintaining robust generalization to unseen antibodies and antigens. In further analysis of the selective attention mechanism, we found that MambaAAI successfully uncovers critical epitope and paratope regions in the SARS-CoV-2 antibody examples. It is believed that MambaAAI holds great potential to discover lead candidates targeting specific antigens at a lower burden.

Anticancer peptides (ACPs) have emerged as a promising class of therapeutic molecules due to their low toxicity and high specificity. However, experimentally identifying ACPs is labor-intensive and low-throughput, making virtual screening with intelligent models an attractive alternative. Despite recent progress, the number of ACP samples available in existing databases remains limited, which may cause predictive models to overfit the labeled information and exhibit suboptimal generalization. To address this issue, we propose a semi-supervised prediction framework with bio-inspired data augmentation for ACP identification, called AugACP. Specifically, we incorporate semantic knowledge into masked Transformer pretraining to guide ACP sequence augmentation and provide more reliable supervised signals. Based on the augmented data, we adopt a semi-supervised learning strategy that selects high-confidence pseudo-labels from augmented peptides to improve prediction performance. In the computational experiments, AugACP outperforms state-of-the-art baseline methods under various experimental configurations, particularly in generalizing to unseen oligopeptides. Moreover, t-SNE visualizations indicate that AugACP learns more discriminative embeddings for separating ACPs from non-ACPs, while external validation on novel ACPs also achieves an F1-score of 0.9272. These results highlight the predictive power of AugACP and its potential value in promoting anticancer drug discovery.