Illustrative model example
Separate the time to reach a state from the time to leave it.
In a two-state model, A moves to B with probability 0.20 per lag and B returns to A with probability 0.10. The same compact model answers three useful questions: how much B is populated, how long it takes to reach B, and how long B persists before returning. Those answers point to different design levers.
- 2/3Stationary population in B
The model spends two-thirds of its long-run time in B.
- 5 lagsMean passage from A to B
Calculated as 1 / 0.20.
- 10 lagsMean passage from B to A
Calculated as 1 / 0.10.
- What goes in · Supply the transition matrix and lag: A stays A with probability 0.80; B stays B with probability 0.90. The remaining probabilities describe movement between the states.
- What comes out · B’s stationary population is 0.20 / (0.20 + 0.10) = 2/3. Mean passage takes 5 lags from A to B and 10 in the reverse direction. Population alone would conceal this directional timing.
- The decision it supports · If B is the useful conformation, ask whether the program needs to reach it sooner, retain it longer, or engage it within the available window. Established MSM tools calculate these quantities too. BioTwin makes them available alongside other views of the supplied landscape so subsequent intervention questions can reuse the same qualified model.
Markov state modeling is established. Tools such as deeptime already provide estimation and kinetic analysis. BioTwin’s implemented contribution is a queryable landscape object with several scientific views; deeper production coupling of the retained trajectory-fitting research remains in development.
| Question | Established approach | BioTwin operation | Practical value |
|---|---|---|---|
| What is populated? | An adequately estimated state model supports stationary-population analysis. | Queries weights and observables on the supplied landscape. | Ask a supported occupancy question from the prepared representation. |
| How does it move? | Kinetic-analysis libraries calculate passage times, committors and transition paths. | Exposes passage, committor, relaxation and flux operations alongside landscape views. | Inspect which timescale or transition a proposed intervention would need to change. |
| When must work be repeated? | The model must be reassessed when sampling, conditions or the scientific question exceed its scope. | The same scientific requirement applies to each reused landscape. | Reuse is valuable when the model remains adequate for the next question. |
This describes implemented operations and a research integration direction. It is not a measured speedup over deeptime or a replacement for new physical evidence.
From a state picture to a mechanism question
A rare state may be reached frequently but disappear quickly, or be difficult to reach and long-lived once formed. Those alternatives suggest different intervention questions. Supported queries include mean first-passage times, committors and relaxation estimates.
Those questions stay attached to one landscape with explicit states and transitions. The same prepared scientific object can support a different design decision without rebuilding the explanation around disconnected outputs.
Prepare a model once. Ask several supported questions.
A supplied landscape holds states, weights and transitions. BioTwin can query its represented populations, mean first-passage times, committors and flux without regenerating atomistic frames for each query. That creates a practical route to repeated analysis on a compact model. Source sampling, fitting and validation still carry their original cost, and a new condition or poorly represented state may require new work.
Connect a timescale to a design choice
A program could compare a state’s lifetime with capture or reversal kinetics, or ask whether a reduced model preserves the slow process of interest. These are specific compositions to evaluate, not automatic consequences of fitting a landscape.
- Capture and reversal · Investigate what a transient state would require from engagement kinetics. Capture and reversal
- Slow-mode preservation · Review the scope of temporal model reduction. Slow-mode preservation