FISIK

FISIK infers in situ molecular interaction kinetics, including association and dissociation rates, from live-cell single-molecule imaging data with substoichiometric labeling.

Key Features:

• Mathematical modeling: Simulates molecular movement and interactions while representing the biological system and the specifics of the data-acquisition setup.
• Parameter estimation: Estimates unknown model parameters such as association and dissociation rates by fitting the model to experimental single-molecule tracking data.
• Indirect inference: Uses the method of indirect inference to calibrate the stochastic model to data.
• Validation with simulated trajectories: Validated using simulated trajectories of diffusing, interacting molecules across a broad spectrum of model parameters.
• Account for experimental limitations: Incorporates effects of resolution limitations, tracking errors, and potential model–system mismatches in analyses.
• Sensitivity characterization: Quantifies how accuracy and precision depend on labeled fraction and the extent of tracking errors.
• Data integration: Combines dynamic sparse single-molecule imaging data with nearly complete population oligomer distribution data to improve parameter inference when labeled fraction is low.

Scientific Applications:

• In situ kinetic quantification: Derives association and dissociation rates of molecular interactions within live cells from single-molecule imaging.
• Analysis under substoichiometric labeling: Infers interaction kinetics when only a small fraction of molecules are labeled.
• Assessment of experimental impacts: Evaluates how resolution limits, tracking errors, and labeling fraction affect kinetic parameter estimates.

Methodology:

Constructs a stochastic mathematical model that simulates diffusing interacting molecules and the imaging acquisition, fits the model to single-molecule tracking data to estimate parameters (association/dissociation rates) using indirect inference, and validates performance with simulated trajectories incorporating resolution limitations, tracking errors, and model–system mismatches.