TinderMIX: Time-dose integrated modelling of toxicogenomics data.

Area of Science:

• Toxicogenomics
• Computational Biology
• Transcriptomics

Background:

• Omics technologies are crucial for toxicology, but analyzing dose and time effects simultaneously is challenging.
• Current methods often analyze molecular alterations at single time points, missing dynamic, dose-dependent responses.
• Identifying gene alteration patterns correlated with both dose and time is a key challenge in toxicogenomic studies.

Purpose of the Study:

• To propose TinderMIX, a novel approach for simultaneously modeling time and dose effects on the transcriptome.
• To investigate the dynamic course of molecular alterations in response to substance exposure.
• To identify genes exhibiting time-dependent and dose-dependent responses.

Main Methods:

• TinderMIX fits integrated time and dose models to gene log fold-changes.
• It selects the optimal model for each gene and computes a time and dose effect map.
• A user-defined threshold identifies responsive genes and their point of departure.

Main Results:

• TinderMIX was applied to analyze cyclosporin A and thioacetamide from the Open TG-GATEs dataset.
• The method identified dynamic, dose-dependent mechanisms of action for both drugs.
• Analysis highlighted the recapitulation of toxicity potential and dynamic mechanisms via integrated time and dose departure points.

Conclusions:

• TinderMIX effectively models the dynamic interplay of time and dose on gene expression.
• The approach provides insights into compound-specific toxicity mechanisms.
• Integrated time and dose analysis enhances understanding of toxicological responses.