AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

molecular-dynamics

Runs and analyzes molecular dynamics simulations with OpenMM and MDAnalysis. Sets up protein/small molecule systems, defines force fields, runs energy minimization and production MD, and analyzes trajectories (RMSD, RMSF, contact maps, free energy surfaces). For structural biology, drug binding, and biophysics.

Original source
K-Dense-AI
License
MIT
Pinned revision
5f2bc0341354fd5590aef76a533ada1e1e5b4e62

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read k-dense-ai--molecular-dynamics at revision 5f2bc0341354fd5590aef76a533ada1e1e5b4e62, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 3

references/system_preparation.md

System preparation and force-field provenance

Reviewed 2026-10-01. OpenMM 8.6.1 is the executed simulation release. PDBFixer 1.12.0 was exercised on a water/ion fixture. OpenFF Toolkit 0.18.0 APIs below were reviewed in official documentation; OpenFF parameterization is illustrative and was not executed in this audit.

PDBFixer: preserve scientific decisions

PDBFixer can propose missing residues from SEQRES and replace modified residues, but neither operation establishes the correct experimental construct. Its removeHeterogens(True) keeps water and removes ligands, ions and cofactors; do not use it as general protein-ligand preparation. This conservative example adds missing atoms/hydrogens without silently rebuilding entire loops or deleting heterogens. Inspect retained nonstandard residues and parameterize them separately. Use conda install -c conda-forge pdbfixer in a separate environment if needed.

from pdbfixer import PDBFixer
from openmm.app import PDBFile

def fix_pdb(input_pdb, output_pdb, ph=7.0):
    fixer = PDBFixer(filename=input_pdb)
    fixer.findMissingResidues()
    print("Missing residue proposals:", fixer.missingResidues)
    fixer.missingResidues = {}  # Explicit policy: do not invent absent loops/termini.
    fixer.findNonstandardResidues()
    print("Nonstandard residues retained:", fixer.nonstandardResidues)
    # Review before opting into replaceNonstandardResidues or heterogen deletion.
    fixer.findMissingAtoms()
    fixer.addMissingAtoms()
    fixer.addMissingHydrogens(ph)
    with open(output_pdb, "w") as handle:
        PDBFile.writeFile(fixer.topology, fixer.positions, handle, keepIds=True)
    return output_pdb

Hydrogens assigned at a pH are a model choice, not a pKa calculation. Inspect histidines, catalytic groups, metal binding, disulfides and termini manually. Run Modeller.addHydrogens(..., pH=...) using the intended force field and consistent pH afterward; retained unknown ligands still need explicit chemical preparation.

OpenFF ligand parameters

The official Toolkit installation route uses conda-forge/Mamba and chemistry backends. Although Toolkit 0.18.0 and Interchange 0.5.1 are on PyPI, their metadata does not declare the complete dependency stack; a bare pip install is insufficient. For example, mamba create -n md-openff -c conda-forge openff-toolkit openff-interchange is an illustrative environment command; check the solved force-field/charge backend dependencies before running it.

Sage 2.3.0 (openff-2.3.0.offxml) is the stable small-molecule release reviewed here; OpenFF 3.0 alpha force fields are research prereleases. Sage 2.3 uses AshGC charges and needs the corresponding OpenFF NAGL implementation and model. Record and validate those versions; do not silently substitute older AM1-BCC charges. Installing/resolving these optional models was outside this small smoke audit.

from openff.toolkit import Molecule, ForceField as OFFForceField

def parameterize_ligand(smiles, ff_name="openff-2.3.0.offxml"):
    """Illustrative ligand-only parameterization; requires the charge backend/model."""
    mol = Molecule.from_smiles(smiles)  # Undefined stereochemistry raises by default.
    mol.generate_conformers(n_conformers=1)
    off_ff = OFFForceField(ff_name)
    return off_ff.create_interchange(mol.to_topology())

The SMILES must encode the intended stereoisomer, tautomer and formal charge. A generated conformer is not the bound pose. Map every ligand atom back to the complex before merging through a supported template/system construction route. Preserve nonbonded conventions, water/ion parameters, exceptions and charge totals; do not concatenate independent OpenMM Systems or confuse OpenFF with GAFF2.

Official sources