Towards Reliable Pro-Apoptotic Peptide Prediction: Debiasing and Confidence-Aware Fusion

<p>This repository contains the foundational datasets and structural validation files for the <strong>Confidence-Gated Fusion Network (CGFN)</strong>, a deep learning framework for pro-apoptotic peptide prediction.</p> <p><strong>1. CGFN_Sequence_Dataset.zip:</strong><br>Includes the raw positive pro-apoptotic peptides, the negative candidate pool, and the rigorous 5-level Functional Mutational Lineage partition matrix (<code>CGFN_Dataset_Partitions.csv</code>) used to ensure zero data leakage during cross-validation.</p> <p><strong>2. CGFN_AlphaFold3_Structures.zip:</strong><br>Contains the downstream structural hierarchy screening results of candidate peptides. It includes the AlphaFold3 ranking metrics and representative 3D complex structures (.pdb) depicting interactions with apoptotic regulatory proteins (e.g., Bax and Bfl-1).</p>