pydangle top2018 full: Geometric measurements and classifications for 12,125 high-quality protein chains with residue filtering based on main chain and side chain metrics

Per-residue backbone geometry (phi, psi, omega, tau), sidechain torsion (chi1), Ramachandran classifications, and DSSP secondary structure for 2,405,322 quality-filtered protein residues from 12,125 chains in 11,843 high-resolution, low-redundancy protein structures. Computed using pydangle-biopython v0.5.1 on "ersatz" full-structure PDB files that preserve multi-chain context for correct DSSP assignments. Source data from the Richardson Lab top2018 dataset (Williams et al., 2021, doi:10.1002/pro.4239), full-filtered, 70% homology, 60% completeness. See README.md for full methodology and ISSUES.md for preparation details.本数据集包含11843个高分辨率、低冗余度蛋白质结构中12125条多肽链的2405322个经质量过滤的蛋白质残基的相关信息，具体包括每个残基的主链几何结构（φ、ψ、ω、τ）、侧链二面角（chi1）、拉马钱德兰（Ramachandran）构象分类以及DSSP二级结构。本数据集通过pydangle-biopython v0.5.1，对保留多链上下文以确保DSSP分配准确的"ersatz（替代式）"完整结构PDB文件计算得到。其源数据来自Richardson实验室的top2018数据集（Williams等，2021，DOI:10.1002/pro.4239），该数据集经全量过滤，设置同源性阈值为70%、序列完整性阈值为60%。完整实验方法详见README.md文件，数据制备细节可查阅ISSUES.md文件。