lqqhwei/HyperDriver: HyperDriver v5.0.0

HyperDriver v5.0.0 Overview This is the Official PyTorch Implementation of the paper: "Key Driver Protein Identification Based on Spectral Energy Proxy and Multi-scale Spatio-temporal Hypergraph". This release provides the complete source code to reproduce all experiments reported in the manuscript, demonstrating that HyperDriver reduces control energy by ≈10^2× compared to traditional baselines. Key Features & Modules The repository implements the unified framework described in the paper: Module 1: Dynamic Interaction Weight Reconstruction. Module 2: Multi-scale Spatio-temporal Hypergraph. Module 3: Key Driver Protein Identification. Reproducibility This release includes a hierarchical workflow for full reproduction: One-Click Pipeline: Run python main.py for end-to-end preprocessing, training, and evaluation. Physical Verification: Exact Lyapunov energy calculation scripts in energy_case_study/. Case Studies: Candidate selector for identifying hidden drivers in driver_case_study/. Environment Python: 3.12+ PyTorch: 2.5+ Hardware: Tested on NVIDIA GeForce RTX 4060 Laptop GPU. Citation If you use this code or our results, please cite: @article{HyperDriver2025, title = {Multi-Scale Spatio-Temporal Hypergraph Modeling of Dynamic PPI Networks and Key Driver Protein Identification via Adaptive Gating and Minimum-Energy Control}, author = {Qiangqiang Li}, note = {Manuscript submitted, under review}, year = {2025} } License MIT License.