Modular HPC orchestration platform for molecular docking and dynamics — unified CLI for fpocket, AutoGrid4, AutoDock-GPU and GROMACS 2025 over SLURM.
翻译准备中
正在加载完整 README。
查看完整 README ↗ChemLink is a modular orchestration platform for computational chemistry, developed at the Computational Chemistry and Biology Laboratory of Universidad del Norte. It grew out of the need to turn an underused HPC infrastructure (several nodes with NVIDIA GPUs and no coordination between them) into an automated, traceable and reproducible scientific computing environment.
The platform automates two experimental workflows. The molecular docking pipeline coordinates active-site detection (fpocket), structure preparation (Meeko/AutoGrid4) and GPU conformational search (AutoDock-GPU). The molecular dynamics pipeline runs a complete GROMACS simulation, from topology preparation to trajectory analysis, for six types of biological systems. Both pipelines share a single CLI and run on a single node or distributed through SLURM job arrays.
The project also covers the HPC infrastructure it runs on: a cluster with passwordless SSH between nodes, NFS shared storage on OpenMediaVault, and continuous monitoring with Prometheus and Grafana. Validation benchmarks show experiment setup time dropping from 45 minutes to under 5, a 100% success rate on 1,000-ligand docking campaigns, and a speedup of up to 2.2× with three GPUs in multi-node mode.
The full documentation (installation, every command and option, the pipelines, HPC deployment and troubleshooting) is at athalea.co/chemlink/docs.
| Document | Description |
|---|---|
| Report.md | Main project report |
| Installation.md | Installation, development and deployment guide |
| Development.md | Technical development details |
| Entropy.md | Binding energy with IE/C2 entropy correction |
curl -fsSL https://raw.githubusercontent.com/athalea-software/chemlink/main/install.sh | bash
Installs ChemLink in /opt/chemlink together with the bio Conda environment.
curl -fsSL https://raw.githubusercontent.com/athalea-software/chemlink/main/install.sh | bash -s -- --full
Builds and installs fpocket, AutoGrid4, AutoDock4, AutoDock-GPU and GROMACS 2025.4. Requires CUDA Toolkit 12.x, OpenMPI and cmake. Expected time: 45–90 min.
| Option | Description |
|---|---|
| (no options) | Installs the Python layer and Conda environments |
--full | Builds all scientific tools, including GROMACS |
--with-gromacs | Builds GROMACS only |
--dir PATH | Installation directory (default: /opt/chemlink) |
--version TAG | Git branch/tag to install (default: main) |
--skip-conda | Skips creating the Conda environments |
chemlink # interactive launcher (TUI)
chemlink doctor # checks the environment, GPU and dependencies
chemlink docking --help # docking pipeline options
chemlink dynamic --help # dynamics pipeline options
chemlink docking peptide --help # peptide-protein docking (LightDock + PRODIGY)
No path option is required. Each command reads its inputs from the directory it runs in and writes its results there:
| Input | Looked up in |
|---|---|
| Receptors | ./receptors/, otherwise the *.pdb files in the directory |
| Ligands | ./ligands/, otherwise the *.sdf/*.mol2/*.mol/*.pdbqt files in the directory |
| Peptides | ./peptides/, otherwise the *.sdf files in the directory |
| Dynamics | protein, peptide, nucleic acid or ligand, detected from file content |
| Results | ./results/<pipeline>/ (-o DIR for another base) |
cd my_project # 1hsg.pdb, ligand.sdf
chemlink docking full # → results/docking/run_…
chemlink dynamic pligand # → results/dynamics/run_pligand_…
chemlink dynamic analyze # latest run in results/dynamics
Executable paths are set with environment variables (FPOCKET_BIN,
AUTOGRID4_BIN, AUTODOCK_GPU_BIN, VINA_BIN, PRODIGY_BIN) or found on PATH; chemlink doctor shows which ones it found.
The interface (TUI, progress bars, logs and reports) runs on any machine: it only
needs Python ≥ 3.9 with rich and prompt_toolkit. Recent Ubuntu/Debian releases
don't let pip install into the system Python, so use a virtual environment
(or sudo apt install python3-rich python3-prompt-toolkit):
python3 -m venv .venv && source .venv/bin/activate
pip install "rich>=13" prompt_toolkit
./chemlink demo # simulated docking + dynamics run (runs nothing)
./chemlink demo dynamics --fail # includes a failure to show the error report
./chemlink demo --plain # what it looks like in a SLURM log file
./chemlink # interactive launcher (TUI)
./chemlink doctor # tools, GPU and paths diagnostics
Global options (accepted anywhere on the command line): --plain (log lines instead
of animated bars; turned on automatically outside a terminal), --no-color, -v/-q, --debug.
以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗