StableProt: Decontaminated Benchmark Holdouts and Prospective Validation Datasets for Protein Stability (Tm) and OGT

StableProt Archive StableProt is a structure-aware dual-pathway deep learning framework for predicting protein melting temperature (Tm) and organismal optimal growth temperature (OGT), with observation-level uncertainty calibration. This archive contains the complete publication-ready manuscript bundle, supplementary materials, figures (PNG and vector SVG), tables, and sanitized prospective experimental validation datasets for industrial biocatalysts: 87 unique enzymes across marine carrageenases and lipases. Contents Manuscript & Supplementary Text Complete authoritative manuscript drafts in: Markdown LaTeX HTML Main Figures & Tables Figures 1–7: high-resolution PNG and vector SVG Tables 1–5 Supplementary Materials Supplementary Figures S1–S3 Supplementary Tables S1–S5 Detailed hyperparameter sweeps Prospective Biocatalysis Validation Data Complete amino acid sequences Predicted Tm and OGT Calibrated 95% confidence intervals Reference literature endpoints Provided in Supplementary Data 1 Evaluation Holdouts Benchmark evaluation suites for: ProThermDB FireProtDB BacDive BRENDA FLIP Meltome Code & Resources Resource Link GitHub Repository https://github.com/Bibhuprasadbehera/StableProt Pre-Trained Weights (Hugging Face) https://huggingface.co/bibhu111/StableProt