Evaluating adaptive stress response gene signatures using transcriptomics.

Area of Science:

• Computational biology
• Toxicogenomics
• Molecular toxicology

Background:

• Stress response pathways (SRPs) are crucial for cellular defense against chemical insults.
• Excessive SRP activation can lead to adverse cellular outcomes, necessitating robust assessment methods.
• Transcriptomic data offers a powerful tool for evaluating cellular responses to chemical perturbations.

Purpose of the Study:

• To develop and evaluate a computational approach for assessing SRP activity using transcriptomic data.
• To compare the performance of novel consensus SRP signatures against existing published signatures.
• To establish a reliable method for characterizing SRP activity in response to new chemical entities.

Main Methods:

• Extracted published gene signatures for six key SRPs (DDR, UPR, HSR, HPX, MTL, OSR) from MSigDB.
• Constructed consensus SRP signatures using a gene-frequency approach.
• Utilized gene set enrichment analysis (GSEA) and receiver-operator characteristic (ROC) analysis on reference transcriptomic datasets.

Main Results:

• Consensus SRP signatures demonstrated comparable or superior performance to published signatures for 4 out of 6 SRPs.
• Achieved high Area Under the Curve (AUC) values for consensus signatures, notably 1.00 for DDR and HPX.
• Successfully classified perturbagens with high accuracy (78% and 88% at first and second rank, respectively) using transcriptomic profile matching.

Conclusions:

• The developed computational approach effectively characterizes SRP activity from transcriptomic data.
• Consensus SRP signatures offer a promising alternative or complement to existing signatures for toxicological assessments.
• This method holds potential for evaluating the SRP activity of novel chemicals, aiding in safety profiling.