Artificial Intelligence-Enhanced Analysis of Genomic DNA Visualized with Nanoparticle-Tagged Peptides under Electron...

Area of Science:

• Biotechnology
• Microscopy
• Nanotechnology

Background:

• Electron microscopy (EM) offers high resolution for DNA visualization, but novel staining agents are lacking.
• Existing methods for DNA staining in EM have limitations in universality and simplicity.
• Advancements in DNA visualization are crucial for understanding genomic structures and functions.

Purpose of the Study:

• To introduce a novel, non-toxic, and universal staining method for DNA visualization using gold nanoparticle-tagged peptides for EM.
• To develop an artificial intelligence (AI)-enabled pipeline for analyzing EM images of DNA molecules.
• To enhance the accuracy and scope of DNA visualization and dimensional analysis in EM.

Main Methods:

• Developed a staining technique utilizing gold nanoparticle-tagged peptides for all naturally occurring DNA types.
• Implemented an AI pipeline for identifying, classifying (size, shape, conformation), and extracting dimensional features of DNA from EM images.
• Validated the AI pipeline's performance on linear (salmon sperm DNA) and circular (M13 phage DNA) molecules.

Main Results:

• The gold nanoparticle-peptide method enables universal, non-toxic DNA staining for EM, allowing sequence-specific and multi-conformational visualization.
• The AI pipeline significantly improved the accuracy of quantifying DNA molecules (number, length, diameter) in EM images.
• The AI pipeline demonstrated robustness against variations in raw EM images, enhancing data reliability.

Conclusions:

• The proposed method offers a simple, universal, and non-toxic approach to DNA staining for EM, expanding visualization capabilities.
• The AI pipeline provides a powerful tool for automated analysis of DNA in EM images, yielding crucial dimensional information.
• This integrated approach advances DNA imaging, analysis, and understanding at the genomic scale.