DENA

DENA detects N^6-methyladenosine (m^6A) modifications in RNA from Oxford Nanopore Technologies direct RNA sequencing data.

Key Features:

• Deep learning model: Implements a neural network-based deep learning approach to identify m^6A from sequencing signals.
• Training data: Trained on in vivo transcript sequencing data from Arabidopsis thaliana.
• Direct RNA sequencing compatibility: Operates on direct RNA sequencing data generated by Oxford Nanopore Technologies.
• Avoids synthetic RNA artifacts: Uses real biological samples to mitigate signal distortion caused by saturated m^6A residues in synthetic RNA.
• miCLIP concordance: Recapitulates approximately 90% of m^6A sites previously detected by miCLIP in Arabidopsis.
• SCARLET consistency: Produces modification rates consistent with SCARLET measurements in human samples.
• Single-nucleotide profiling: Evaluates single-nucleotide m^6A profiles, including analyses of mtb and fip37-4 m^6A-deficient mutants.
• Cross-species applicability: Demonstrates utility across plant and human datasets through validation and comparative analyses.

Scientific Applications:

• m^6A site identification: Detects N^6-methyladenosine sites at single-nucleotide resolution from Nanopore direct RNA data.
• Modification rate quantification: Estimates modification rates that align with SCARLET measurements in human samples.
• Mutant transcriptome analysis: Profiles m^6A landscapes in m^6A-deficient mutants such as mtb and fip37-4.
• Cross-method validation: Enables benchmarking of Nanopore-based m^6A detection against miCLIP and SCARLET.
• Comparative epitranscriptomics: Supports comparative analyses of m^6A methylation across species and biological contexts.

Methodology:

Neural network deep learning models were trained on in vivo Oxford Nanopore direct RNA sequencing data from Arabidopsis thaliana and applied to Nanopore direct RNA reads to detect and quantify single-nucleotide m^6A sites, with validation against miCLIP and SCARLET.