Probabilistic Reconstruction of Truncated Particle Trajectories on a Closed Surface

Yunjiao Lu, Pierre Hodara, Charles Kervrann, Alain Trubuil

In the field of cell biology, tracking the movement of individual proteins is essential for understanding fundamental processes like cell wall synthesis. However, a persistent challenge in live-cell imaging is the "truncation" of trajectories. When particles move on a 3D surface—such as the membrane of a rod-shaped bacterium—they frequently disappear from the 2D microscope's field of view, creating fragmented "tracklets."

The Challenge of the "Dark Side"

A classic example of this problem is the movement of MreB proteins in Bacillus subtilis. These proteins move along the inner membrane of the cylindrical cell. Using Total Internal Reflection Fluorescence Microscopy (TIRFM), researchers can only observe the "front" of the cell (the Region of Observation, or ROO). As a protein rotates around the circumference, it enters a "hidden" zone on the back of the cell, only to reappear later.

Standard tracking algorithms often fail to link these re-entrance events, especially when particles spend significant time in the unobserved region or when particle density is high. Without accurate reconstruction, researchers cannot determine the total lifetime of a protein or the number of full rotations it completes.

Modeling the "Dark Side" of the Cell

To address this, the authors developed a mathematical framework that treats trajectory reconstruction as a global optimization problem. The core of the approach involves "unwrapping" the cylindrical surface into a 2D plane where the top and bottom edges are topologically connected.

Microscopy data (a) and the corresponding geometric model (b) showing how 3D cylindrical trajectories are unwrapped into 2D tracklets with hidden regions.

The motion is modeled as Brownian motion with drift. By calculating the "first passage time" of a particle through the unobserved region, the framework can determine the likelihood that a particle exiting the ROO at one point is the same one entering it later at another.

A Global Optimization Approach

Unlike "greedy" algorithms that match segments sequentially, this method uses Integer Linear Programming (ILP) to find the global maximum likelihood configuration for an entire movie simultaneously. This prevents local errors from cascading through the dataset.

The framework is also remarkably data-driven. It includes built-in estimators for key parameters such as drift velocity ($v$), variance ($\sigma$), and particle birth/death rates. This removes the need for manual parameter tuning, making the tool more accessible for biological researchers.

Validation and Biological Impact

The method was rigorously tested using both synthetic and real-world datasets:

  • Synthetic Data: In simulations mimicking B. subtilis dynamics, the algorithm achieved an Adjusted Rand Index (ARI) above 0.9 for moderate densities. Even under high-traffic conditions (100+ tracklets), the ARI remained robust at over 0.7.
  • Biological Application: When applied to TIRFM movies of MreB aggregates, the tool successfully connected fragmented segments, allowing the authors to estimate the distribution of rotations—a key proxy for bacterial cell wall synthesis activity.

Conclusion

By providing a rigorous probabilistic solution to the problem of truncated trajectories, this work transforms fragmented data into continuous molecular "stories." This framework represents a significant advancement in quantitative bio-imaging, offering a more complete and accurate picture of protein dynamics on curved biological surfaces.

Summary updated

Abstract

Investigation of dynamic processes in cell biology very often relies on the observation in two dimensions of three-dimensional (3D) biological processes. Consequently, the data are partial and statistical methods and models are required to recover the parameters describing the dynamical processes. In the case of molecules moving over the 3D surface, such as proteins on walls of bacteria cell, a large portion of the 3D surface is not observed in two-dimensional (2D)-time microscopy. It follows that biomolecules may disappear for a period of time in a region of interest, and then reappear later. Assuming Brownian motion with drift, we address the mathematical problem of the reconstruction of biomolecules trajectories on a cylindrical surface. A subregion of the cylinder is typically recorded during the observation period, and biomolecules may appear or disappear in any place of the 3D surface. The performance of the method is demonstrated on simulated particle trajectories that mimic MreB protein dynamics observed in 2D time-lapse fluorescence microscopy in rod-shaped bacteria.
BibTeX
@article{lu2021probabilistic,
  author = {Yunjiao Lu and Pierre Hodara and Charles Kervrann and Alain Trubuil},
  title = {{Probabilistic Reconstruction of Truncated Particle Trajectories on a Closed Surface}},
  journal = {Multiscale Model. Simul.},
  year = {2021},
  doi = {10.1137/20m1333742},
}
Figure