CropStabDB: a protein stability resource and Engineering Atlas of orthologue-supported substitutions for 30 plant proteomes

CropStabDB: a protein stability resource and Engineering Atlas of orthologue-supported substitutions for 30 plant proteomes Version 2.1. CropStabDB is a protein stability resource for 30 plant species - 29 crops and Arabidopsis thaliana, spanning 12 families and 1,425,716 proteins. Each protein carries a CropStab score for folding propensity and six interpretable components computed from 64 sequence features. 850,949 proteins also carry a per-residue AlphaFold confidence layer, served as data; the remaining 574,767 carry a sequence-based disorder layer, and every protein records which layer it has. The Engineering Atlas lists 2,097,576 candidate substitutions in 51,440 proteins, 785,622 of them at an engineering confidence score of 0.70 or above. Every candidate is a residue that more stable orthologues already carry at the aligned position, reported with its support, a confidence score, and an 11-residue sequence-context string that locates it without counting. This record contains: - the release database, with per-protein scores, components and features, and the Engineering Atlas- the released classifier and the component parameter files, including the full feature dictionary for all 64 features- the frozen two-task benchmark with its evaluation script and baseline files- the validation outputs: agreement with measured folding free energies, three held-out proteomes, leave-one-species-out and orthogroup-blocked cross-validation- the provenance record identifying every input proteome by assembly, provider, file name and md5 No proteome FASTA is redistributed here. The 30 proteomes are third-party, the providers' data-use terms differ, and no right to republish them is claimed. 03_provenance/species_provenance.tsv identifies each input file exactly, and regenerate_processed.py rebuilds the cleaned sequences from a proteome you download yourself and verifies them against the database. Web interface: https://cropstab.orgSource code: https://github.com/Israfil-Hossen/CropStabDB Supersedes version 2 (10.5281/zenodo.21412425) and version 1.0.0 (10.5281/zenodo.20551450). Values reported in those versions were superseded; use this one.