Claude Molecular Dynamics Session Demo

You

Run a 5 ps simulation on 1EBY_dimer_ligand_BEB.pdb, poll every 7 minutes and visualise the results.

Claude

check_structure.py

Blocked before upload 1EBY_dimer_ligand_BEB.pdb is not supported as it stands. The BEA369 inhibitor sits inside the receptor as 48 HETATM records, and there is no path to parameterize a non-protein molecule embedded in the protein file — prep would fail on a missing template, or silently drop the ligand.

Everything else is clean: 2 chains, 99 residues each, no altlocs, no insertion codes, no missing-residue gaps. The 95 crystallographic waters get replaced with fresh TIP3P, which is expected.

The fix: delete the HETATM records and supply BEB as a separate posed 3D SDF. I can do that here — it needs bond-order perception, so I would install RDKit first.

You

Do that, then try again.

Claude

pip install rdkit RCSB · BEB.cif check_structure.py

Bond orders come from the PDB Chemical Component Dictionary rather than a guess. All 48 atom names matched the CCD entry exactly.

Extracted	48 heavy atoms, 93 bonds
Hydrogens	40 added, MMFF-relaxed with heavy atoms fixed
Formula	C38H40N2O8 — matches CCD

The docked pose is untouched — only hydrogen positions were relaxed.

Re-checked · supported One warning: ligand protonation is used as-is. BEA369 is neutral at pH 7, so this is benign.

Claude

Ready to submit — confirm before I start.

Setting	Value
Input	1EBY_receptor.pdb + BEB.sdf
Simulation time	5 ps — 2500 steps @ 0.002 ps
Ensemble	NVT — minimize → equilibrate → produce
Force field	ff19SB / TIP3P (default)
Platform	CUDA (default)
Estimated	~4 min · ~4 credits

Submit this run? nothing uploaded yet

You

Submit.

Claude

prepare_local_file_upload submit_job get_job_status

Job ID	21173bb3-c386-4a31-b0d0-d7476cc469c6
Params	simulation_time_ps: 5
Status	running

polling every 7 min9m 36s

Claude

get_result_url trajectory_viewer.py

Success. 5 ps NVT, ff19SB/TIP3P, OpenMM 8.0 on CUDA.

Metric	Value
Temperature	300.73 ± 1.74 K
Cα RMSD	0.59 Å final · 0.62 max
Mean Cα RMSF	0.35 Å
PE drift	0.20 %
System	198 res · 3134 protein + 88 ligand atoms

Energy bounded, temperature held, structure intact. The Asp25/Asp25′ catalytic dyad holds the inhibitor diol at ~100 % occupancy across all 125 frames.

Viewer is on the right — switch between the ligand pocket and the chain interface to see contacts with live distances.

At 5 ps this shows the system is stable and correctly built. It is far too short for binding conclusions.

Example session, lightly condensed. The viewer alongside is the real output of job 21173bb3 — 125 frames of a 5 ps run, fully interactive and running entirely in this page.

Loading trajectory…