Recent advances in multimodal foundation model-enabled peptide screening and optimization for smart biomaterials and functional tissue engineering.

Peptides are important bioactive molecules for smart biomaterials and functional tissue engineering because they can provide targeting, antimicrobial, adhesive, immunomodulatory, and differentiation-regulating functions. However, conventional peptide discovery remains limited by high experimental cost, incomplete exploration of sequence space, and the difficulty of balancing bioactivity with stability, safety, manufacturability, and material compatibility. These limitations become more pronounced when peptides must retain function after chemical modification, immobilization, hydrogel incorporation, crosslinking, printing, or construct maturation. Recent advances in protein language models and multimodal foundation models provide new opportunities for peptide screening, generation, property prediction, and material-aware prioritization. This review summarizes foundation model-assisted strategies for peptide representation learning, candidate retrieval, de novo generation, interaction and property prediction, multi-objective optimization, benchmarking, developability assessment, and staged material-level validation. We also discuss representative applications in antimicrobial biomaterials, targeting systems, immunomodulatory materials, peptide-functionalized hydrogels, hydrogel bioinks, and three-dimensional bioprinted constructs. Overall, foundation model-assisted peptide design should be viewed as a candidate-prioritization framework rather than a substitute for experimental validation. Its future value will depend on standardized material-context data, transparent benchmarking, and validation across solution, material, biofabrication, and construct-level settings.