Gene expression classification using epigenetic features and DNA sequence composition in the human embryonic stem...

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Existing studies show correlations between epigenetic factors and gene expression.
• Quantitative models for classifying gene expression based on epigenetics are lacking.

Purpose of the Study:

• To develop a machine learning model for accurately classifying highly and lowly expressed genes using epigenetic data.
• To improve predictive accuracy by integrating various epigenetic features.

Main Methods:

• A machine learning approach combining histone modifications, DNA methylation, DNA accessibility, transcription factors, and trinucleotide composition.
• Support Vector Machines (SVM) were utilized for classification.
• The model was developed and tested on a human embryonic stem cell line (H1).

Main Results:

• The inclusion of epigenetic features significantly enhanced predictive accuracy.
• The best model achieved 95.96% accuracy (10-fold cross-validation) and 95.58% (independent dataset).
• Matthews correlation coefficients reached 0.92 for both testing methods.

Conclusions:

• The developed model effectively predicts gene expression levels using genetic and epigenetic data.
• This provides a valuable tool for researchers lacking gene expression data.
• A web server, GECES, is available for easy access to the analysis method.