A convolutional neural network-based regression model to infer the epigenetic crosstalk responsible for CG...

Area of Science:

• Epigenetics
• Genomics
• Computational Biology

Background:

• Epigenetic modifications like DNA methylation and histone modifications interact to influence genomic patterns.
• The precise nature of epigenetic crosstalk regulating CG methylation is not fully understood, particularly in limited cell samples like mammalian oocytes.
• Existing machine learning methods primarily focus on DNA sequence rather than epigenetic interactions.

Purpose of the Study:

• To develop a machine learning model to predict CG methylation patterns based on epigenetic features, excluding DNA sequence.
• To investigate epigenetic crosstalk involved in regulating CG methylation in mammalian oocytes.

Main Methods:

• Development of a machine learning model named epiNet.
• Utilizing histone modifications (e.g., H3K36me3, H3K4me3) as input features to predict CG methylation patterns.
• Application of the model to mouse oocytes, mutant oocytes, and cross-species prediction for human oocytes.

Main Results:

• epiNet successfully predicted CG methylation patterns using histone modifications.
• Identified significant epigenetic crosstalk between H3K36me3, H3K4me3, and CG methylation in mouse oocytes.
• The model accurately predicted altered CG methylation in mutant oocytes and showed cross-species applicability to human oocytes, suggesting broader relevance.

Conclusions:

• The study provides novel insights into the epigenetic crosstalk governing CG methylation patterns in mammalian oocytes.
• The developed epiNet model offers a valuable tool to complement and guide experimental design in epigenetics research.
• Findings highlight potential epigenetic crosstalk relevant to various cell types beyond oocytes.