Exploring regulatory elements in low-methylated regions for gene expression prediction.

Area of Science:

• Epigenetics and Genomics
• Computational Biology

Background:

• Low-methylated regions (LMRs) are increasingly recognized as potential active distal regulatory elements.
• Understanding the link between LMRs and gene expression regulation is crucial for deciphering cellular function.

Purpose of the Study:

• To develop and evaluate a predictive model for gene expression directional change.
• To investigate the role of transcription factor binding sites within LMRs in gene expression regulation.

Main Methods:

• A penalized logistic regression model was developed to predict gene expression changes.
• The model utilized computationally analyzed transcription factor binding sites within LMRs.
• Whole genome bisulphite sequencing and RNA-seq data from adipose-derived stem cells and iPSCs were used for evaluation.

Main Results:

• The model achieved a 10-fold cross-validated AUC of 0.88 for predicting expression directional change in differentially expressed genes with LMRs in intergenic and/or genebody regions.
• For differentially expressed genes with LMRs exclusively in intergenic regions, the model yielded an AUC of 0.84.

Conclusions:

• Transcription factor binding sites in LMRs are strong predictors of gene expression directional change.
• The proposed computational model effectively links epigenetic marks (LMRs) to gene expression dynamics.