NAMD runs classical molecular dynamics: atoms move under a force field rather than being
treated with quantum mechanics, so you can simulate thousands of atoms, such as a protein
in a box of water. ChemCompute builds the simulation for you. You say what goes in the box,
and it:
Go to NAMD → Submit (/namd/submit).
Fastest possible first run: the form opens with ubiquitin (PDB entry
1UBQ) in water.
Press Submit without changing anything to see the whole workflow. It is a big system
(about 35,000 water molecules), so expect it to take a while; see
Keep it small for a quicker test.
Under System Type, pick one. Changing the type clears the molecules you've entered.
| Option on the form | Use it for |
|---|---|
| PDB (e.g. protein) (default) | A protein or other structure from the Protein Data Bank, by its 4-character ID |
| Predefined molecule | Molecules already set up for you: sodium, chloride, pentane, neopentane, isopentane, propane |
| Small molecule (search for any molecule) | Any small organic molecule, found by name. Only C, H, O, N, S and P, and about 3–20 atoms |
| NaCl crystal | A block of rock-salt crystal (no solvent) |
| Upload my own prebuilt system (.psf + .pdb) | Researcher accounts only; see below |
Each row in the molecule table is one kind of molecule. Count is how many copies are
packed into the box. Use Add another … for more rows and Delete to remove one.
With no rows at all you get a box of pure water.
1UBQ. The structure is downloaded from the RCSB Proteinbenzene and press Search. Wait for thefound and check the picture before you submit. If you press3x3x3. Each unit cell is 5.64 Å and two atoms wide, so 5x5x50ChemCompute cleans up a PDB entry before building the system:
Could not generate parameters for: …. Use a structureThe form fills these in with sensible defaults.
| Setting | Default | What it does |
|---|---|---|
| Box size X / Y / Z (Å) | 75 each | The region your molecules are packed into |
| Padding (Å) | 15 (0 for a crystal) | Extra solvent on every side. Simulation cell = box + 2 × padding, so the defaults give a 105 Å cube |
| Solvent | Explicit Waters | Fill the cell with water, or No Solvent. The number in brackets is an estimate of how many water molecules that adds. (Reset all values switches this to No Solvent.) |
| Ensemble | NVT (Constant Volume) | Or NPT (Constant Pressure), which adds Pressure (bar) (default 1.01325) and Cell fluctuation |
| Temperature (K) | 310 | Whole numbers |
| Run minimization / Minimization steps | on, 10000 | Relaxes clashes before dynamics. Leave it on for a new system |
| Dynamics / Dynamics steps | on, 10000 | Number of MD steps |
| Time step (fs) | 2 | 10,000 steps × 2 fs = 20 ps of simulation |
For an air/water interface or a membrane slab under NPT, choose Constant area (NPAT)
under Cell fluctuation; an isotropic piston squeezes the interface itself. Write the
pressure tensor adds the Pxx, Pyy and Pzz components to the energy plot and CSV, which is
what a surface tension is calculated from.
You don't choose these; ChemCompute sets them:
Run time grows with the number of atoms, and with the defaults nearly all of them are
water. For a small molecule, try a box of 20 Å with 10 Å padding (a 40 Å cell, about 2,000
waters). The run-time estimate under the resources section updates as you change
settings.
Press Use Restart File and enter the Job Number to use for Restart. The molecules,
box and solvent all come from that job, minimization is turned off, and the step count
becomes Additional dynamics steps. You can only restart from your own jobs, and only
once the earlier job has written output.
Choose the cluster, cores and time limit (see Clusters, cores and time limits),
optionally a Project, then press Submit. On success you'll see
Your job #N has been submitted to <cluster> and be taken to the results page.
If your account has GPU access, NAMD can run on a GPU on Expanse, Bridges-2 or Delta.
A GPU speeds up large solvated systems a lot. With a GPU chosen, the cores setting is
ignored.
Open the job from your Dashboard or NAMD → Results. The page opens on
Output while the job runs, and moves to Summary when it finishes.
| Tab | What's there |
|---|---|
| Summary | Status, your settings, and a diagnosis if something went wrong |
| Output | NAMD's log, live while the job runs. It's only available while the job is running |
| Trajectory (HTMoL) | Plays the trajectory in the browser. Choose under Trajectory to view: protein + ligand + ions without water, all atoms including water (fewer frames), or just the ligand. Press Load PDB, then Load Trajectory. Highlight ligand colours it magenta |
| Trajectory (nglview) | An alternative viewer in a notebook. Large jobs can freeze it; Restart Server recovers |
| Analysis | Plot Energy (total, potential and kinetic energy, temperature, pressure, density and the individual terms), Download Energy as CSV, radial distribution functions, and a Ramachandran plot for peptides |
Download Output File gets a zip of everything. Wait about a minute after the status
says COMPLETE.
Radial distribution functions (RDFs): pick a First atom type and a Second atom
type (e.g. water oxygen OH2 with itself) and press Calculate. Each RDF runs as a
small job of its own, taking from 30 seconds to 15 minutes. It covers half the box length
and the second half of the trajectory.
Ramachandran plot: for a protein or peptide, Show Ramachandran plot gives the φ/ψ map or
the angles over time, the share of frames in each region, and a CSV of every angle.
Researcher accounts (every verified .edu account) see a fifth system type,
Upload my own prebuilt system.
Your files are used exactly as given: nothing is padded, solvated or neutralized.
.psf and a .pdb..coor, .vel and .xsc restart files. These are ignored if.par, .prm, .str).conf files. They are generated from the settings on the form.See NAMD and TINKER errors. The two you're most likely to
meet:
FATAL ERROR: ERROR(S) IN THE CONFIGURATION FILE.Atoms moving too fast; simulation has become unstable: make sure minimization is