DeFiNe: an optimisation-based method for robust disentangling of filamentous networks.

Area of Science:

• Multidisciplinary science
• Network analysis
• Image processing

Background:

• Thread-like structures, such as protein polymers, root systems, and galaxy filaments, are common across various scales.
• Network formalisms are often used to study these structures, but existing methods have limitations.
• Current network link representations often capture only partial filaments, potentially leading to inaccurate analyses.

Purpose of the Study:

• To develop a fully automated and robust method for detecting filaments with consistent intensities and angles within a network.
• To overcome the limitations of existing filament detection methods that require extensive parameter tuning and treat all filaments uniformly.
• To facilitate the automated analysis of diverse filamentous systems.

Main Methods:

• An optimization-based approach was developed to detect filaments in network representations.
• The method identifies filaments based on consistent intensity and angular properties.
• The accuracy of the approach was validated using contrived, biological, and cosmic structures.

Main Results:

• The proposed approach accurately detects filaments in various types of networks.
• It enables powerful automated analysis of individual actin filaments within their network context.
• The tool, named "DeFiNe" (Decomposition of Filaments in Networks), was made publicly available as open-source software.

Conclusions:

• The developed optimization-based approach provides a robust and automated solution for filament detection in networks.
• DeFiNe facilitates the decomposition of complex networks into individual filaments for detailed study.
• This tool enhances the ability to analyze diverse filamentous systems, including biological and cosmic structures.