deepmodeling / DeePDih

Deep Potential driven molecular dihedral scan toolkit

DeePDih

Deep Potential driven dihedral scan toolkit.

Installation

Requirements

• Python 3.8 or later (3.10 recommended)
• PyTorch 2
• DeePMD-kit (at least 3.0.0a0)
• networkx
• RDKit
• ASE
• Geometric
• DPData
• OpenMM (for adding bias and reading Gromacs topology)
• tblite (for semi-empirical level geometry optimization)
• Psi4 (for high-level quantum mechanical single point energy)

Install requirements using conda

Users can install the required packages by running the following command:

conda install -c conda-forge rdkit geometric tblite-python psi4 openmm

If you want to use Psi4 to compute RESP charges, you also need to install Psi4 with the following command:

conda install -c psi4 psi4

Install requirements from PyPI

pip install ase dpdata networkx

If you want to use Gromacs topology, you also need to install ambertools, acpype and parmed.

conda install -c conda-forge ambertools
pip install acpype parmed

Install requirements from source

DeePMd-kit

Please look at DeePMD-kit's installation guide

Geometric-jit

Accelerate Geometric using Numba and JIT. Please look at this repository.

Numba is also needed if you want to use the JIT version of Geometric.

conda install -c conda-forge numba

Install DeePDih module from source

pip install .