01

Characterize
not to conclude—
but to ask
the next question.

A biophysical system is characterized sufficiently to determine how it should be perturbed next.

CLOSED-LOOP BIOPHYSICAL INTERROGATION
THE SHIFT02
NOT

The strongest excitation.

BUT

The most informative excitation.

The optimal input makes competing mechanisms produce measurably different responses.

THE LEARNING LOOP03
Observe→Infer→Excite→Update
Dn → p(M, θ | Dn) → u*n+1 → yn+1 → Dn+1

Every measurement updates the model. The model chooses the next measurement.

ADAPTIVE MEASUREMENT04

Measure where the models disagree.

OPTIMAL PERTURBATION05

Perturb the system to expose hidden dynamics.

ligandtemperaturepHforceelectric fieldlightflowshear
OPTIMAL CONTROL06

Interrogate first.
Steer second.

The input can move beyond information gain and drive a desired molecular state. Pulse shape, phase, frequency and timing become model-informed controls.

Observation becomes intervention without losing the connection to the physical model.

APPLICATION · SPR07

The Oracle does not request more data. It requests the right experiment.

koff uncertain
Extend dissociation.
Transport coupled
Change concentration and flow rate.
Rebinding suspected
Use pulses of different duration.
Models ambiguous
Add competition or regeneration steps.
APPLICATION · SAXS08

Which condition pushes competing structures furthest apart?

Input space
q-range · time · temperature · concentration · pH · ligand · flow
Model question
Which parameters remain unidentifiable?
Experimental question
Which next condition maximizes structural discrimination?
THE EXPERIMENTAL ORACLE09

Characterization is not the end of the experiment.

It generates the next optimal question to the physical system.

Data → Parameter & Structure Oracle → Uncertainty Map → Experiment Oracle → New Data

Selected foundations

Optimal experimental design in systems biology ↗ Active learning for biological experimental design ↗ Adaptive NMR experiment design ↗ Optimal control pulses for NMR ↗ Real-time adaptive experimentation ↗