DNA-m6A calling and integrated long-read epigenetic and genetic analysis with fibertools

Area of Science:

• Genomics
• Epigenetics
• Bioinformatics

Background:

• Long-read DNA sequencing offers single-molecule resolution for genetic and epigenetic studies.
• Detecting DNA N-methyladenine (DNA-m6A) and integrating genetic/epigenetic data is computationally challenging.
• Existing tools lack comprehensive support for coprocessing diverse long-read sequencing data.

Purpose of the Study:

• To develop a fast and accurate computational tool for DNA-m6A detection using long-read sequencing.
• To enable integrated analysis of genetic and epigenetic data from single DNA molecules.
• To address the limitations of current tools in handling complex genomic and epigenomic information.

Main Methods:

• Development of fibertools, a toolkit featuring a semisupervised convolutional neural network.
• Utilized Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) sequencing platforms.
• Implemented algorithms for DNA-m6A identification and coprocessing of genetic/epigenetic data.

Main Results:

• Achieved >90% precision and recall for DNA-m6A identification on long DNA molecules (>20 kb).
• Demonstrated an approximately 1000-fold speed improvement compared to existing methods.
• Successfully integrated genetic and epigenetic data, enabling seamless coordinate system conversion.

Conclusions:

• Fibertools provides a state-of-the-art solution for DNA-m6A detection and integrated genomic analysis.
• The toolkit enhances the analysis of structurally and somatically variable genomic regions.
• Facilitates advanced genetic and epigenetic studies using long-read sequencing data.