MethylationToActivity

MethylationToActivity infers promoter activity landscapes from genome-wide DNA methylation to predict H3K4me3 and H3K27ac enrichment and enable gene-level regulatory interpretation.

Key Features:

• Deep-learning framework: Employs convolutional neural networks (CNNs) to recognize complex patterns in high-dimensional DNA methylation data.
• Promoter activity inference: Infers promoter activities from DNA methylation profiles to link methylomes to individual gene regulatory impacts.
• Epigenetic marker focus: Predicts enrichment of H3K4me3 and H3K27ac as indicators of active promoter states.
• Public dataset training: Processes and trains on publicly available DNA methylation datasets to simulate real-world scenarios.

Scientific Applications:

• Tumor detection and classification: Extends DNA methylome biomarker utility by elucidating gene-level biological impacts relevant to tumor detection, subtyping, and classification.
• Cancer research: Validated in pediatric and adult cancers, including solid tumors and hematologic malignancies.

Methodology:

Processes publicly available DNA methylation datasets and trains convolutional neural networks on methylation patterns to predict promoter activities reflected by H3K4me3 and H3K27ac enrichment.