Rebuilding the full range of shapes a protein moves through, so supported low-population states come into view.
A disordered protein is a population, not a single structure.
That population spans dominant disordered conformations and rare states that appear only briefly. No single structure can represent the target, and simulation alone cannot determine which states the protein actually occupies.
Peptone combines HDX-MS measurements in solution with OPES enhanced sampling and physiological reweighting to recover the ensemble. NMR and EPR add orthogonal local and long-range restraints that test whether the modeled populations agree with experiment.
Enhanced sampling expands the conformational search; experimental agreement narrows it. Elastic AWS and NVIDIA infrastructure supplies the capacity required for simulation, reweighting, and inference while keeping each calculation tied to the measurements that determine what should be tested next.
00:00 00:00
Accelerated computing at Peptone
A short film on the infrastructure behind Peptone's ensemble-first, experimentally grounded Product Engine.
Many aligned conformations represent the theoretical population. No single trace is the protein.
OPES multiT
One simulation visits more shapes.
A known bias helps the simulation cross barriers. Because the bias is recorded, each frame can be reweighted to physiological temperature.
A known bias helps one simulation visit common and rare shapes. Reweighting then recovers their populations at physiological temperature.
HDX-MS
Peptide uptake sets the weights.
Deuterium uptake shows which peptide regions stay protected or exposed in solution.
NMR
Local signals check transient structure.
Chemical shifts and relaxation test where short helices appear and how often.
EPR
Distances check long-range contacts.
Spin-label distributions test how often distant regions meet.
HDX-MS weights the ensemble against peptide protection. NMR and EPR then check local structure and long-range contacts.
Experimental weights reshape basin populations without erasing the rare conformations already sampled by physics.
A compact folded subensemble remains part of the wider weighted population. Its transient helix and long-range contact agree with physics, HDX-MS, NMR, and EPR.
Ensemble gate
A simulated ensemble becomes useful only when its weighted populations reproduce independent measurements across timescales.
Simulation becomes an experimentally weighted ensemble
An OPES ensemble and solution measurements merge into physical plausibility, experimental agreement, and rare-state resolution gates. Supported populations remain in the ensemble, while disagreement returns the model to sampling and reweighting.
Generated ensemble
Enhanced sampling explores folded and disordered conformations, then reweighting estimates their populations at physiological conditions.
ModeledExperimental observables
Complementary measurements weight local exchange, secondary structure, long-range contacts, and dynamics without forcing one static structure.
MeasuredConvergent evidence
Ensemble and experiment agreeAre the states thermodynamically coherent?
Sampling and reweighting must support a plausible population under physiological conditions.
Do orthogonal observables converge?
HDX-MS, NMR, and EPR test the same population through complementary solution measurements.
Do the weights resolve rare states?
A supported minor population must remain physically sampled while improving agreement across experiments.
Evidence decision
Support the ensemble
Convergent measurements support a population map that includes both dominant and rare conformations.
Return to sampling
Disagreement redirects sampling, experimental restraints, or population weights before interpretation.
Refine sampling or weights
