iModEst: disentangling -omic impacts on gene expression variation across genes and tissues

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Gene expression is influenced by various regulatory factors like microRNA, long non-coding RNA (lncRNA), transcription factors (TFs), methylation, copy number variation (CNV), and single-nucleotide polymorphisms (SNPs).
• The relative importance of these regulatory mechanisms at the individual gene and tissue level remains largely uncharacterized.

Purpose of the Study:

• To develop and present the integrative Models of Estimated gene expression (iModEst) tool.
• To quantify the relative contribution of different regulators to gene expression across 16,000 genes and 21 tissues using The Cancer Genome Atlas (TCGA) data.

Main Methods:

• Derived predictive models of gene expression using tumor data from TCGA.
• Tested the predictive accuracy of these models in both cancerous and tumor-adjacent tissues.
• Analyzed relationships between various regulators (miRNA, lncRNA, TFs, methylation, CNV, SNPs) and gene expression.

Main Results:

• iModEst models explained up to 70% of gene expression variance for 43% of genes in both tumor and adjacent tissues.
• Transcription factor (TF) expression was the most consistent predictor of gene expression across both tissue types.
• Methylation predictive models showed poor transferability from tumor to adjacent tissues.
• Identified novel and confirmed known regulator-gene-expression relationships (e.g., CNV-FGFR2, SNP-TP63).

Conclusions:

• iModEst provides a comprehensive atlas of regulator-gene-tissue expression relationships.
• TF expression is a key determinant of gene expression in both tumor and adjacent tissues.
• The tool offers insights into the complex interplay between genetic/epigenetic regulators and gene expression patterns.