HelixSeed: An α-Helical Seed Resource for MaSIF-Seed-Driven de novo Binder Design

This dataset comprehensively integrates nearly 400,000 α-helical fragments from multiple mammalian species, constituting an ultra-large-scale candidate seed library with exceptional biological diversity. It provides an immensely broad structural search space for protein binders design.The greatest advantage of this database lies in its out-of-the-box, pre-computed nature.

The molecular surface meshes and complex physicochemical feature fingerprints for all fragments in the library have been standardized and pre-computed in advance. This resolves the long-standing bottleneck of lacking a publicly available seed library in surface-driven protein design workflows. Researchers no longer need to expend massive computational resources and extensive time on underlying data cleaning and feature extraction.