Related Experiment Video
Updated: Jun 22, 2026
13:06
Visualization of miniSOG Tagged DNA Repair Proteins in Combination with Electron Spectroscopic Imaging ESI
Published on: September 24, 2015
10.2K
Image recovery from unknown network mechanisms for DNA sequencing-based microscopy.
David Fernandez Bonet1, Ian T Hoffecker1
1Science for Life Laboratory, Department of Gene Technology, KTH Royal Institute of Technology, Tomtebodavägen 23a 171 65, Solna, Sweden. ithof@kth.se.
Nanoscale
|April 20, 2023
Summary
This study introduces a novel graph-based method for reconstructing molecular networks using imaging-by-sequencing. The technique enhances spatial localization accuracy and robustness for diverse molecular imaging applications.
Area of Science:
- Molecular Biology
- Bioinformatics
- Computational Imaging
Background:
- Conventional optical imaging faces limitations in micro- and nanoscale resolution.
- Imaging-by-sequencing methods leverage DNA molecules to record proximity-dependent associations, enabling network reconstruction via sequencing.
- Reconstruction strategies for these molecular networks are an open challenge, particularly regarding accuracy, noise robustness, and scalability.
Purpose of the Study:
- To develop a computational framework for reconstructing diverse molecular networks from imaging-by-sequencing data.
- To address the open problem of determining optimal reconstruction strategies for spatial localization, noise robustness, and scalability.
- To create a unified reconstruction approach applicable to various molecular network generation mechanisms.
Main Methods:
- A graph-based technique is employed for network reconstruction in 2D and 3D.
- Unsupervised sampling using random walks is utilized to capture local and global network structure, ensuring robustness with minimal prior assumptions.
- A two-stage dimensionality reduction process, involving structural discovery and manifold learning, is used to recover network images.
Main Results:
- The proposed method reconstructs diverse molecular network classes without prior knowledge of their generation mechanisms.
- The approach demonstrates robustness to noise and achieves high spatial localization accuracy.
- The staged dimensionality reduction significantly reduces computational complexity, leading to fast and accurate performance.
Conclusions:
- The developed graph-based method offers a unified and efficient framework for reconstructing molecular networks from imaging-by-sequencing data.
- This approach enhances the capabilities of molecular imaging by improving spatial resolution and data analysis.
- The method provides a versatile solution for analyzing complex molecular interaction networks across various biological systems.
