DEEPAligner: Deep encoding of pathways to align epigenetic signatures.

Area of Science:

• Bioinformatics
• Computational Biology
• Epigenetics

Background:

• Differential DNA methylation influences biological pathway regulation.
• Pathways possess methylation signatures distinguishing biological subtypes.
• Next Generation Sequencing (NGS) enables genome-wide epigenetic analysis, but computational methods for pathway comparison are needed.

Purpose of the Study:

• To develop a novel computational method for aligning epigenetic signatures across biological pathways.
• To enable comparison of pathway networks for identifying conserved methylation patterns.

Main Methods:

• Introduction of DEEPAligner (Deep Encoded Epigenetic Pathway Aligner), a novel alignment method.
• Utilizing a deep embedding framework to derive epigenetic signatures from pathways.
• Aligning signatures based on functional consistency and local topological similarity.

Main Results:

• Experiments on four cancer datasets demonstrated conserved epigenetic signatures.
• Identified signatures specific to cancer subtypes and across different cancers.
• DEEPAligner revealed conserved epigenetic patterns in cancer.

Conclusions:

• The deep embedding framework generates coherent signatures aligned for biological and structural orthology.
• DEEPAligner achieves more topologically and functionally consistent alignments than state-of-the-art methods.
• The method provides a robust approach for analyzing epigenetic pathway networks.