KnotResolver: tracking self-intersecting filaments in microscopy using directed graphs.

Area of Science:

• Biophysics
• Computational Biology
• Cell Biology

Background:

• Traditional computational tools struggle to quantify microtubule dynamics in gliding assays when filaments are highly curved or self-intersecting.
• Existing methods are limited to analyzing non-intersecting, rod-like filaments, hindering the study of complex biological structures.

Purpose of the Study:

• To develop and present KnotResolver, a novel computational image-analysis pipeline for tracking highly curved, self-intersecting looped filaments (knots) in microscopy time series.
• To overcome the limitations of current tracking tools in analyzing complex microtubule dynamics.

Main Methods:

• The pipeline integrates filament segmentation and cross-over identification using a directed graph representation.
• Nodes in the graph represent cross-overs, and edges represent the connecting paths, allowing for complex filament reconstruction.
• Graphs are mapped back to contours, minimizing distance to a reference for sub-pixel accuracy and noise robustness.

Main Results:

• KnotResolver successfully tracks image time series of highly curved, self-intersecting looped filaments by resolving cross-overs.
• The method achieves sub-pixel accuracy in contour detection and demonstrates robustness to noise.
• Demonstrated utility in quantifying flagella-like curvature dynamics and wave-like oscillations of clamped microtubules in gliding assays.

Conclusions:

• KnotResolver provides a robust solution for analyzing complex microtubule dynamics in gliding assays, particularly for self-intersecting and highly curved filaments.
• The open-source availability of the MATLAB-based code facilitates broader application in biophysical and cell biology research.
• This tool enhances the quantification of dynamic biological structures, opening new avenues for research in motor-driven transport and cytoskeletal dynamics.