Skip to content
Discovery Stage 10 · MD Simulation

Run production-scale molecular dynamics without managing GPUs

Most small biotech teams don't have the GPU infrastructure for MD - and shouldn't need it. Submit a protein-ligand complex. The engine handles force field selection, solvation, equilibration, production, and analysis. Identify unstable candidates before committing to lab validation.

“Run a 10ns MD simulation of this protein-ligand complex and calculate MM-GBSA binding free energy.”

GROMACS
2023.3 GPU
MM-GBSA
Binding free energy
AlphaFlow
Dynamics fallback
Async
Results streamed
The funnel

How it works

01

Submit a PDB ID or complex

Provide a protein PDB ID, duration, and temperature. The audit_system pre-flight classifier detects membrane proteins, metal sites, and heme clusters - routing complex cases to the appropriate pipeline or AlphaFlow.

02

Production MD on GPU

GROMACS runs on Azure GPUs. Soluble proteins use AMBER99SB-ILDN. Membrane proteins use CHARMM36m via packmol-memgen. Metalloprotein parameters via MCPB.py. Jobs run async - minutes to hours depending on system size.

03

Structured results returned

RMSD convergence, RMSF flexibility, radius of gyration, hydrogen bond analysis, equilibration assessment (temperature, pressure, density, energy). MM-GBSA binding free energy. Full trajectory data with interactive charts.

Proof

GROMACS 2023.3 on GPU. OpenFF Sage 2.x force fields. PACKMOL-Memgen for membrane systems.

audit_system pre-flight: OPM membrane detection, metal site coordination geometry, heme/Fe-S cluster identification. Routes complex cases to AlphaFlow automatically.

parameterize_metal: QM→FF bridge via MCPB.py for metalloprotein MD. Async jobs with email notifications on completion via Resend.

Use this when you need to

Confirm binding stability over time - not just a static score

Evaluate protein flexibility and conformational dynamics

Prioritize candidates before committing to experiments

Run membrane protein simulations without CHARMM-GUI setup

Research preview

Validate candidates before the lab - not after

GROMACS on GPU. MM-GBSA binding energy. Identify unstable candidates computationally.