Predicting gene expression changes upon epigenomic drug treatment.

Area of Science:

• Epigenetics and Cancer Genomics
• Computational Biology and Bioinformatics

Background:

• Tumors exhibit altered epigenetic modifications like DNA methylation and histone modifications, crucial for tumor progression.
• Epigenetic drugs, including histone deacetylase inhibitors (HDACi) and DNA methyltransferase inhibitors (DNMTi), are explored for cancer therapy.
• A key challenge is the lack of genomic specificity, leading to unpredictable transcriptional changes and variable drug responses.

Purpose of the Study:

• To assess the predictability of locus-specific gene expression changes following histone deacetylase inhibitor (HDACi) treatment.
• To utilize machine learning models integrating pre-treatment transcriptome and epigenomic data for predicting treatment effects.

Main Methods:

• Employed machine learning to predict gene expression alterations upon HDACi treatment.
• Utilized pre-treatment transcriptome and epigenome profiles alongside post-treatment transcriptome data for model training and validation.

Main Results:

• Achieved high accuracy (ROC up to 0.89) in distinguishing upregulated versus downregulated genes post-treatment in HCT116 and RH4 cell lines.
• Demonstrated generalizability of the predictive model across different cell lines, indicating robustness.

Conclusions:

• Presents the first assessment of predictability for genome-wide transcriptomic changes induced by HDACi.
• Highlights the need for comprehensive omics data from clinical trials to evaluate the clinical applicability of predictive models for epigenetic drugs.