AGZArank source data

This Zenodo record contains the source data and supporting files used to reproduce the benchmark evaluation reported in the AGZArank manuscript. It includes benchmark source PDB structures, processed benchmark inputs in PDB and JSONL formats, baseline reproduction scripts and outputs, Graphinity enrichment calculation files, and ranking result tables used to compute the reported Top-k recovery and enrichment metrics. The associated GitHub repository provides the AGZArank code, inference scripts, and usage instructions. Large source PDB files are archived here due to file size constraints. The proprietary AGZArank training dataset is not included because it contains Arlan Biotech-generated structural data.