nvmolkit-usage - Use nvMolKit for batched GPU cheminformatics
Write code using the installed nvMolKit Python API for batched GPU-accelerated RDKit-style cheminformatics operations.
Tags
Updated: 2026-10-03cheminformaticsGPU computingRDKitCUDAmolecular fingerprintsconformer optimizationsimilarity searchsubstructure search
Capabilities
Typical Outputs
What this skill does
- Generate Morgan fingerprints
- Compute similarity matrices
- Embed molecular conformers
- Optimize molecular geometries
- Compute RMSD and TFD
- Cluster molecular fingerprints
- Search molecular substructures
- Configure GPU execution
- Chain GPU tensor results
- Diagnose nvMolKit calls
Inputs
- RDKit molecule lists
- Fingerprint arrays
- Operation parameters
- GPU hardware configuration
- PyTorch or NumPy pipeline requirements
- CUDA stream settings
Outputs
- Python code
- GPU tensors
- NumPy arrays
- RDKit conformer coordinates
- Optimization energies and convergence flags
- Similarity, RMSD, or TFD matrices
- Clustering results
- Substructure match results
Requirements
- NVIDIA GPU with compute capability 7.0+
- CUDA 12.6+ compatible driver
- CUDA-enabled PyTorch
- Installed nvMolKit package
- Installed RDKit package
- Optuna package for autotuning