CoGTEx: Unscaled system-level coexpression estimation from GTEx data forecast novel functional gene partners.

Area of Science:

• Genomics
• Bioinformatics
• Systems Biology

Background:

• Coexpression analysis is crucial for understanding biological pathways, gene regulation, and human diseases.
• Current coexpression estimations often rely on tissue-level standardization, potentially overlooking valuable information from overall gene expression levels.
• There is a need for coexpression methods that incorporate diverse cell types and account for non-uniform data across tissues.

Purpose of the Study:

• To challenge the assumption that variance is more important than mean gene expression levels in coexpression analysis.
• To develop and validate a "system-level" coexpression estimation method that does not rely on tissue standardization.
• To provide a comprehensive resource for exploring both tissue-level and system-level coexpression data from human GTEx data.

Main Methods:

• GTEx v8 expression data underwent global normalization, batch processing, and filtering.
• Principal Component Analysis (PCA), clustering, and t-distributed Stochastic Neighbor Embedding (tSNE) were used to define 42 distinct tissue clusters.
• Coexpression was calculated for 33,445 genes across 42 tissue clusters, using sampling to avoid tissue overrepresentation and repeated 20 times for robust estimation. Pearson, Spearman, and G-statistic metrics were computed at both system-level (TPM scale) and tissue-level (z-score scale).

Main Results:

• Tissue-level coexpression estimations were validated against existing databases.
• System-level coexpression estimations demonstrated differences from tissue-level estimations, with both revealing valuable information about biological pathways.
• Coexpression estimations were found to be associated with transcriptional regulation, and the CoGTEx resource was presented for analyzing human adult tissue coexpression data.

Conclusions:

• System-level coexpression represents a novel and valuable metric for generating biological hypotheses and predictions.
• The CoGTEx resource offers a unique platform for viewing, comparing, and downloading both system- and tissue-level coexpression estimations.
• This study highlights the importance of considering overall gene expression levels in coexpression analysis for a more comprehensive understanding of biological systems.