Data and model checkpoints for T4E-Miner

This record contains the processed datasets and model checkpoints associated with T4E-Miner, a structure- and function-guided weak-label expansion framework for mining tagatose 4-epimerases. The data archive includes five-fold CD-HIT-based training datasets. The checkpoint archive contains five fold-specific best model checkpoints obtained during cross-validation. The fold 1 checkpoint was used for MGnify inference in this study. The original large-scale public databases and pretrained ESM-C model weights are not included and should be obtained from their original sources.