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/mdanalysis_analysis.md

MDAnalysis Analysis Reference

Reviewed against MDAnalysis 2.10.0 on 2026-10-01. Examples use local topology and trajectory files and are illustrative outside the synthetic regression checks. Atom order, molecule integrity, box vectors and time units must be checked first.

MDAnalysis Universe and AtomGroup

import MDAnalysis as mda

# Load Universe
u = mda.Universe("topology.pdb", "trajectory.dcd")
# or for single structure:
u = mda.Universe("structure.pdb")

# Key attributes
print(u.atoms.n_atoms)          # Total atoms
print(u.residues.n_residues)    # Total residues
print(u.trajectory.n_frames)   # Number of frames
print(u.trajectory.dt)         # Time step in ps
print(u.trajectory.totaltime)  # Stored span, (n_frames - 1) * dt, in ps

Atom Selection Language

MDAnalysis uses a rich selection language:

# Basic selections
protein = u.select_atoms("protein")
backbone = u.select_atoms("backbone")  # CA, N, C, O
calpha = u.select_atoms("protein and name CA")
water = u.select_atoms("resname WAT or resname HOH or resname TIP3")
ligand = u.select_atoms("resname LIG")

# By residue number
region = u.select_atoms("resid 10:50")
specific = u.select_atoms("resid 45 and name CA")

# By proximity
near_ligand = u.select_atoms("protein and around 5.0 resname LIG", updating=True)

# By property
charged = u.select_atoms("resname ARG LYS ASP GLU")
hydrophobic = u.select_atoms("resname ALA VAL LEU ILE PRO PHE TRP MET")

# Boolean combinations
active_site = u.select_atoms("(resid 100 102 145 200) and protein")

# Inverse
not_water = u.select_atoms("not (resname WAT HOH)")

Geometric selections use periodic dimensions by default; updating=True re-evaluates the selection each frame. Verify residue naming (protein uses recognized residue names) and use resindex plus a saved identity map when IDs repeat across chains.

Common Analysis Modules

RMSD and RMSF

from MDAnalysis.analysis import rms, align

# Make molecules whole using trusted bonds before this structural analysis.
# Keep original coordinates/boxes for separate periodic contact calculations.
aligned = u.copy()
aligned.trajectory[0]
align.AlignTraj(aligned, aligned, select='backbone', in_memory=True).run()

# RMSD
R = rms.RMSD(aligned, select='backbone', groupselections=['protein and name CA'], ref_frame=0)
R.run()
# Shape (n_frames, 4): frame, time (ps), backbone RMSD, CA RMSD (Å).
# CA RMSD uses the backbone fit; groupselections are not fitted independently.

# RMSF (per-atom fluctuations)
from MDAnalysis.analysis.rms import RMSF
rmsf = RMSF(aligned.select_atoms('backbone')).run()
# rmsf.results.rmsf: per-atom RMSF values in Angstroms

Radius of Gyration

protein = aligned.select_atoms("protein")  # Must be whole; valid atomic masses required.
rg = []
for ts in aligned.trajectory:
    rg.append(protein.radius_of_gyration())
import numpy as np
print(f"Mean Rg: {np.mean(rg):.2f} Å")

Secondary Structure Analysis

from MDAnalysis.analysis.dssp import DSSP

# DSSP secondary structure assignment per frame
dssp = DSSP(u).run()
# dssp.results.dssp: per-residue per-frame secondary structure codes
# H = helix (including 3_10 and pi), E = strand, - = loop/unordered.
# Requires complete, correctly ordered backbone atoms; not eight-state DSSP.

Hydrogen Bonds

from MDAnalysis.analysis.hydrogenbonds import HydrogenBondAnalysis

# Illustrative BACKBONE-only selection; adapt names to the actual topology.
# Explicit H selection avoids charge-based guessing on PDB files without charges.
# With reliable bonds, prefer donors_sel=None to pair bonded donor and H atoms.
hbonds = HydrogenBondAnalysis(
    u,
    donors_sel="protein and name N",
    hydrogens_sel="protein and name H HN",
    acceptors_sel="protein and name O",
    d_h_cutoff=1.2,          # donor-H distance (Å)
    d_a_cutoff=3.0,          # donor-acceptor distance (Å)
    d_h_a_angle_cutoff=150   # D-H-A angle (degrees)
)
hbonds.run()

counts = hbonds.count_by_time()  # One count per analyzed frame (including zeros).
# The result table has one row per bond occurrence, not one row per frame.
import pandas as pd
df = pd.DataFrame(hbonds.results.hbonds,
                  columns=['frame', 'donor_ix', 'hydrogen_ix', 'acceptor_ix',
                           'DA_dist', 'DHA_angle'])

Principal Component Analysis (PCA)

from MDAnalysis.analysis import pca

# Requires atom types (normally inferred when reading PDB) and matched atom order.
# Fit AND project the same already aligned coordinates. transform() does not align.
pca_analysis = pca.PCA(aligned, select='backbone', align=False).run()

# PC variances
print(pca_analysis.results.variance[:5])  # Raw coordinate variance in Ų.
print(100 * pca_analysis.results.cumulated_variance[:5])  # Cumulative percent.

# Project trajectory onto PCs
projected = pca_analysis.transform(aligned.select_atoms('backbone'), n_components=3)
# Shape: (n_frames, n_components)

Free Energy Surface (FES)

This is a relative, coordinate-dependent -RT log p surface for equilibrated, unbiased sampling at one temperature. Biased/enhanced sampling requires the appropriate reweighting first. Empty bins are unsampled, not measured barriers. Bin size, coordinate Jacobians and autocorrelation affect interpretation; report replicate/block uncertainty. A projected histogram is not a binding free energy.

import numpy as np
import matplotlib.pyplot as plt

def plot_free_energy_surface(x, y, bins=50, T=300, xlabel="PC1", ylabel="PC2",
                              output="fes.png"):
    """
    Compute 2D free energy surface from two order parameters.
    FES = -kT * ln(P(x,y))
    """
    x, y = np.asarray(x, dtype=float), np.asarray(y, dtype=float)
    if x.ndim != 1 or y.shape != x.shape or len(x) < 2:
        raise ValueError("x and y must be equal-length 1D arrays with at least two samples")
    if not np.all(np.isfinite(x)) or not np.all(np.isfinite(y)) or not np.isfinite(T) or T <= 0:
        raise ValueError("Coordinates must be finite and T must be finite and positive")
    R_gas = 0.00831446261815324  # kJ/mol/K (molar gas constant)
    kT = R_gas * T

    # 2D histogram
    H, xedges, yedges = np.histogram2d(x, y, bins=bins, density=True)
    H = H.T

    # Free energy
    H_safe = np.where(H > 0, H, np.nan)
    fes = -kT * np.log(H_safe)
    fes -= np.nanmin(fes)  # Shift minimum to 0

    # Plot
    fig, ax = plt.subplots(figsize=(8, 6))
    im = ax.pcolormesh(xedges, yedges, np.ma.masked_invalid(fes),
                       shading='flat', cmap='RdYlBu_r')
    plt.colorbar(im, ax=ax, label='Free Energy (kJ/mol)')
    ax.set_xlabel(xlabel)
    ax.set_ylabel(ylabel)
    plt.savefig(output, dpi=150, bbox_inches='tight')
    return fig

Trajectory Formats Supported

| Format | Extension | Notes | |--------|-----------|-------| | DCD | .dcd | CHARMM/NAMD binary; widely used | | XTC | .xtc | GROMACS compressed | | TRR | .trr | GROMACS full precision | | NetCDF | .nc | AMBER format | | LAMMPS | .lammpstrj | LAMMPS dump | | HDF5 | .h5md | H5MD standard | | PDB | .pdb | Multi-model PDB |

MDAnalysis Interoperability

# Convert to numpy
positions = u.atoms.positions.copy()  # Snapshot in Å: shape (N, 3)

# Write to PDB
with mda.Writer("frame_10.pdb", u.atoms.n_atoms) as W:
    u.trajectory[10]  # Move to frame 10
    W.write(u.atoms)

# Write a topology with exactly the subset atom order before its trajectory.
protein = u.select_atoms("protein")
protein.write("protein_topology.pdb")
# Write a uniformly sampled trajectory subset, preserving the frame interval.
# DCD cannot represent arbitrary timestamps; separately record the time origin.
with mda.Writer("protein_traj.dcd", protein.n_atoms, dt=u.trajectory.dt) as W:
    for ts in u.trajectory:
        W.write(u.select_atoms("protein"))

# Convert to MDTraj (for compatibility)
# import mdtraj as md
# traj = md.load("trajectory.dcd", top="topology.pdb")

Performance Tips

  • Use in_memory=True for AlignTraj when RAM allows (much faster iteration)
  • Select minimal atoms before analysis to reduce memory/compute
  • Parallel execution: inspect get_supported_backends() for that class; do not assume every analysis or coordinate transformation supports multiprocessing
  • Choose a frame slice with start/stop/step; slicing is not a general streaming guarantee (PCA still builds a coordinate covariance matrix):
# Analyze every 10th frame from frame 100 to 1000
R.run(start=100, stop=1000, step=10)

Official API references