mdtbx
Toolbox for MD simulation
English documentation |
日本語ドキュメント
- Build system quickly
- Run conventional/enhanced MD simulation
- Analyze trajectory
- Calculate Free energy
-
Sampling
-
Free energy calculation
-
Analysis
-
Kinetic analysis
-
Performance
-
Build system
Assumptions
- System build Tools: AMBER, PyMOL, OpenBabel
- Simulation Tools: Gromacs, AMBER, Gaussian16
- ForceField: ff14SB, TIP3P, GAFF2, Lipid21, GLYCAM06-j
Install
# for pixi users
export PIXI_FROZEN=false
pixi install
pixi run pymolrc
ln -s $PWD/.pixi/envs/default/bin/mdtbx $BIN
ln -s $PWD/.pixi/envs/default/bin/pymol $BIN
# for docker users
docker build -t mdtbx .
# for developers
pre-commit install
The PyMOL plugin loader automatically registers cuemol_style,
molstar_style, chimerax_style, and blender_style. See the
plugin guide for commands and galleries.
Update
export PIXI_FROZEN=false; pixi run update
Usage
# for pixi users
pixi run mdtbx ...
pixi run gmx ... # equal to pixi run mdtbx cmd gmx ...
# for docker users
docker run -it --rm mdtbx mdtbx ...
docker run -it --rm mdtbx gmx ... # equal to docker run -it --rm mdtbx mdtbx cmd gmx ...
Agent guide
Run mdtbx skill (or pixi run mdtbx skill from a source checkout) to print
the bundled English Markdown guide for agents. It covers preparation,
simulation, analysis, and recovery, and directs workflows to the maintained
scripts under example/.
pixi run mdtbx skill
pixi run mdtbx skill > MDTBX_SKILL.md
Enhanced workflows
# Build a solution with Packmol-randomized water coordinates.
pixi run mdtbx build_solution -i protein.pdb -o solution \
--boxsize 100 100 100 --water-seed 42
# Retry deterministic placements until a non-clashing orientation is found.
pixi run mdtbx place -1 chain_a.pdb -2 chain_b.pdb -d 30 \
--max-attempts 20 --seed 42 -o placed.pdb
# Match MOL2 atoms by atom index when atom names are duplicated.
pixi run mdtbx mv_crds_mol2 -r reference.mol2 -c coordinates.mol2 \
--match-by index -o updated.mol2
# Choose the final trajectory path and retain the pre-PBC concatenation.
pixi run mdtbx trjcat -n 10 --prefix prd -o fitted.xtc \
--keep-concatenated
# Generate and retain a reusable solvent susceptibility file.
pixi run mdtbx place_solvent -p leap.parm7 -x leap.rst7 \
--solvent-model cSPCE --xvv-output cSPCE_300.xvv \
-o placed_water.pdb
# Preserve independent-trajectory boundaries through tICA, clustering, and MSM.
pixi run mdtbx tica -i replica1.npy replica2.npy \
--lagtime 10 --n-components 3 -o tica.npz
pixi run mdtbx cluster -i tica.npz --n-clusters 100 \
--seed 42 -o clusters.npz
pixi run mdtbx msm -i clusters.npz --lagtime 10 \
--count-mode effective -o msm.npz
# Prepare, run, and analyze a standard GROMACS decoupling calculation.
pixi run mdtbx setup_fep -f example/mdp/solution/prd.mdp \
-p gmx.top -c equilibrated.gro --moltype LIG -o fep
pixi run mdtbx run_fep --path fep
pixi run mdtbx analyze_fep --path fep -b 2000
# Prepare and run PLUMED FEP/REST with the installer-provided gmx_mpi.
pixi run mdtbx setup_fep_rest -f run.mdp -p hybrid.top \
-c equilibrated.gro --replicas 32 -o fep_rest
pixi run mdtbx run_fep --path fep_rest --multidir --replex 1000
pixi run mdtbx analyze_fep_rest --path fep_rest -b 2000
PyMOL AI commands
Inside the configured PyMOL session, claude and codex automatically apply
validated PyMOL code. Completed turns are shared between both backends and
retained only in memory for the current PyMOL session. Requests run one at a
time in submission order.
PyMOL> claude color the ligand red
PyMOL> codex show it as sticks
PyMOL> ai_history
PyMOL> ai_status
PyMOL> ai_clear
Start with the example/ catalog for copy-ready solution,
membrane, Slurm, and analysis scripts. See
example/fep/README.md for standard FEP and FEP/REST,
example/abfe/README.md for ABFE, and
example/msm/README.md for kinetic analysis.
