Enhanced sampling molecular dynamics

What is enhanced sampling molecular dynamics?

Enhanced sampling molecular dynamics is a family of simulation methods that changes how molecular configurations are sampled so slow energy barriers are crossed more often. It can use bias potentials, generalized ensembles, accelerated dynamics, restraints, or related strategies, while methods such as metadynamics, umbrella sampling, and replica-based approaches answer different questions and require method-specific reweighting and convergence evidence.

Choose the enhanced-sampling method from the target observable and known slow coordinates. A poorly chosen collective variable can hide orthogonal barriers, while aggressive bias can distort pathways. Pilot simulations, restraint overlap, bias deposition settings, and independent starts should be planned before production.

ProteinIQ runs well-tempered metadynamics and simulated tempering through OpenMM. Metadynamics applies bias along one to three collective variables and returns bias arrays, collective-variable records, and an estimated free-energy grid. Simulated tempering samples a temperature ladder and returns its temperature and weight log. Both methods save sampling state for continuation.

The connected review workflow separately analyzes an existing compatible trajectory. It does not run enhanced sampling or perform method-specific reweighting. Use external analysis for reweighting and convergence assessment before interpreting populations or free-energy differences; an exported free-energy grid alone does not establish convergence.

When to use enhanced sampling molecular dynamics

• Rare transitions, conformational landscapes, and free-energy profiles
• A supported system for OpenMM sampling, or an existing topology and trajectory with method-specific records for review

Strengths and limitations

Biasing or generalized-ensemble methods can visit states that remain rare in an ordinary trajectory. With suitable collective variables, overlap, and reweighting, they can estimate a free-energy profile for a defined observable.

An omitted slow coordinate can leave apparently well-sampled biased runs trapped. Free-energy estimates depend on bias history, statistical weights, overlap, and convergence; retain these records and compare independent runs before reporting differences.

Enhanced sampling molecular dynamics methods

Metadynamics deposits history-dependent bias along collective variables; umbrella sampling restrains overlapping windows; accelerated and generalized-ensemble methods alter other parts of the Hamiltonian or sampling distribution. Their outputs cannot be interpreted with one generic recipe.

ProteinIQ's OpenMM settings offer well-tempered metadynamics and simulated tempering. Metadynamics supports distance, angle, torsion, and RMSD collective variables. Simulated tempering changes the temperature of one simulation; it is distinct from exchanging configurations between replicas.

Other methods require their own supported engine and protocol. For example, PLUMED integrates many enhanced-sampling and free-energy methods with engines including GROMACS and OpenMM. For a PLUMED-based run, retain its input, bias files, kernels, and reweighting commands as part of the reproducible method. The presence of OpenMM in ProteinIQ does not imply that arbitrary PLUMED protocols can be submitted.

Enhanced sampling molecular dynamics applications

Enhanced sampling molecular dynamics is best suited to rare transitions, conformational landscapes, and free-energy profiles. Match the modeled system, timescale, resolution, and ensemble to the observable rather than choosing a protocol because it produces a longer trajectory or more elaborate figure.

Use simulation as model-based evidence. Connect trajectory observations to experimental data, alternative parameterizations, independent starts, and uncertainty whenever the downstream claim concerns mechanism, affinity, kinetics, stability, or population.

How to run enhanced sampling molecular dynamics online

For the two supported sampling methods, open the OpenMM tool. The Open review workflow button on this page opens a separate post-run trajectory-analysis template.

1. Define objective. Define the target observable and justify collective variables, windows, temperatures, or scaled interactions.
2. Configure sampling. Prepare a supported system in OpenMM, choose dynamics, and set Sampling method to Metadynamics or Simulated tempering. Use a fixed-step integrator. Define the collective variables and bias settings for metadynamics, or the temperature ladder and weight settings for tempering. Other protocols require an external engine.
3. Run and retain records. Sampling starts after preparation stages. Save the trajectory, native system and restart files, and sampling_state.json. Keep the metadynamics bias arrays, collective-variable CSV, free-energy NPY, and grid JSON, or the simulated-tempering TSV log.
4. Analyze the intended ensemble. Apply method-specific reconstruction or reweighting externally where needed. Use the review workflow for compatible topology and trajectory files and complementary structural metrics. Pooled simulated-tempering frames do not represent a single-temperature equilibrium ensemble.
5. Test convergence. Test block convergence, overlap, hysteresis, independent starts, and sensitivity to analysis choices.

How to interpret enhanced sampling molecular dynamics results

Free-energy differences require statistically weighted probabilities in the intended ensemble. A colorful projection of biased frames is not a free-energy surface unless weights, normalization, and convergence are defined.

OpenMM's metadynamics output is a native free-energy estimate in kJ/mol on the configured collective-variable grid. Read metadynamics_grid.json for coordinate definitions and array-axis order. This estimate is distinct from a reweighted distribution of an arbitrary observable. See the native metadynamics documentation for the free-energy output and simulated-tempering documentation for temperature and weight records.

Inspect hidden coordinates, recrossings, window overlap, bias stationarity, independent-run agreement, and uncertainty. Report regions unsupported by sampling instead of smoothing them into apparent states.

How enhanced sampling molecular dynamics works

This template reviews an existing compatible trajectory with RMSD, RMSF, PCA, and clustering. Run supported enhanced sampling separately in OpenMM; prepare method-specific reconstruction or reweighting externally.

1. Prepare review files. Prepare compatible topology and trajectory files from a completed run, with any necessary ensemble reconstruction performed separately.
2. Upload topology. Supply the topology matching the atom ordering in the trajectory.
3. Upload trajectory. Supply the trajectory to review and retain its bias, temperature, and weight records separately.
4. Configure structural analysis. Choose the atom selection and RMSD, RMSF, PCA, or clustering settings for structural review.
5. Interpret structural metrics. Compare structural metrics with independent runs and method-specific convergence evidence; these metrics do not reweight the ensemble.

Inputs and outputs

Check formats before running, then inspect and download the result from every workflow step.

Inputs

Simulation evidence

PDBmmCIFTPRGROXTC

For sampling in OpenMM, a supported prepared system and method settings. For the review workflow, compatible topology and trajectory files; retain bias, collective-variable, temperature, weight, and convergence records separately.

Outputs

Simulation outputs

XTCPDBCSVJSONZIP

OpenMM metadynamics returns bias arrays, collective-variable records, a free-energy estimate and grid metadata; simulated tempering returns temperature and weight records. Both retain sampling continuation state. The separate review workflow returns structural metrics, PCA, and clusters; method-specific reweighting and convergence assessment remain external.