SSBlazer: a genome-wide nucleotide-resolution model for predicting single-strand break sites

Area of Science:

• Genomics and Molecular Biology
• Bioinformatics and Computational Biology
• DNA Repair Mechanisms

Background:

• Single-strand breaks (SSBs) represent a major form of DNA damage crucial for biological processes.
• Existing sequencing-based methods for SSB detection are often costly and not suitable for large-scale genomic studies.
• Understanding SSB distribution is vital for comprehending genome stability and cellular function.

Purpose of the Study:

• To introduce SSBlazer, an innovative deep learning framework for predicting single-strand break (SSB) sites at the nucleotide level.
• To provide a scalable, explainable, and computationally efficient solution for SSB site prediction.
• To enable broader exploration of SSB-related biological questions and applications.

Main Methods:

• Development of a lightweight deep learning model named SSBlazer.
• Training and validation of the model for nucleotide-level SSB site prediction.
• Assessment of the model's generalization capabilities across different species.

Main Results:

• SSBlazer demonstrates high accuracy in predicting SSB sites at the nucleotide level.
• The framework exhibits robust generalization across various species, indicating broad applicability.
• SSBlazer is computationally efficient and scalable for large-scale genomic analyses.

Conclusions:

• SSBlazer offers a significant advancement in DNA damage detection, providing a powerful tool for genomic research.
• The explainable and scalable nature of SSBlazer facilitates numerous unexplored SSB-related applications.
• This deep learning approach enhances our ability to study DNA repair and genome stability.