DeepSAP: An Integrated Generative and Predictive Deep-Learning Framework for Controllable Design of Self-Assembling Peptides.

DeepSAP enables precise, property-guided generation of self-assembling peptides with high reliability and control, outperforming existing models in design tasks.

• Why it matters: Controlling peptide self-assembly is crucial for developing tailored biomaterials, yet current computational methods lack the ability to generate peptides conditionally based on desired properties, limiting design flexibility.
• What they did: The framework combines large-scale peptide pretraining, fine-tuning on SAP data, and a lightweight predictive model, SAPBRF, to generate and screen peptides with targeted physicochemical traits across multiple design scenarios.
• The result: DeepSAP produces peptides that consistently exhibit stable self-assembly behavior verified by molecular dynamics, enabling the tailored design of functional biomaterials with broad application potential.