Fisher information matrix for single molecules with stochastic trajectories

TL;DR

This paper develops a stochastic framework to calculate the Fisher information matrix and Cramér-Rao lower bound for estimating biophysical parameters of single molecules with stochastic trajectories, considering photon detection at discrete times.

Contribution

It introduces a novel method to compute the Fisher information matrix and CRLB for molecular motion models based on photon detection data, extending beyond Gaussian assumptions.

Findings

The methodology accurately predicts the standard deviation of parameter estimates.

Simulation results confirm the CRLB as a reliable benchmark.

The approach applies to complex motion models with discrete photon measurements.

Abstract

Tracking of objects in cellular environments has become a vital tool in molecular cell biology. A particularly important example is single molecule tracking which enables the study of the motion of a molecule in cellular environments and provides quantitative information on the behavior of individual molecules in cellular environments, which were not available before through bulk studies. Here, we consider a dynamical system where the motion of an object is modeled by stochastic differential equations (SDEs), and measurements are the detected photons emitted by the moving fluorescently labeled object, which occur at discrete time points, corresponding to the arrival times of a Poisson process, in contrast to uniform time points which have been commonly used in similar dynamical systems. The measurements are distributed according to optical diffraction theory, and therefore, they would be modeled by different distributions, e.g., a Born and Wolf profile for an out-of-focus molecule. For some special circumstances, Gaussian image models have been proposed. In this paper, we introduce a stochastic framework in which we calculate the maximum likelihood estimates of the biophysical parameters of the molecular interactions, e.g., diffusion and drift coefficients. More importantly, we develop a general framework to calculate the Cram\'er-Rao lower bound (CRLB), given by the inverse of the Fisher information matrix, for the estimation of unknown parameters and use it as a benchmark in the evaluation of the standard deviation of the estimates. There exists no established method, even for Gaussian measurements, to systematically calculate the CRLB for the general motion model that we consider in this paper. We apply the developed methodology to simulated data of a molecule with linear trajectories and show that the standard deviation of the estimates matches well with the square root of the CRLB.