Prediction of human genes and diseases targeted by xenobiotics using predictive toxicogenomic-derived models (PTDMs)

Area of Science:

• Environmental Health Sciences
• Computational Toxicology
• Systems Biology

Background:

• Understanding human exposure to environmental agents is crucial for assessing health risks.
• Existing methods for chemical safety evaluation can be enhanced by integrating complex biological networks.

Purpose of the Study:

• To develop a novel computational-systems toxicology framework, predictive toxicogenomics-derived models (PTDMs).
• To integrate chemical-gene interactions (CGIs), chemical-disease associations (CDAs), and gene-disease associations (GDAs) for hazard profiling.
• To identify data gaps and incorporate network information into chemical safety assessments.

Main Methods:

• Constructed three comprehensive networks: CGI, CDA, and GDA, using data from the Comparative Toxicogenomics Database (CTD).
• Employed a 10-fold cross-validation methodology with 100 simulations to evaluate model performance.
• Achieved high predictive accuracy with areas under the receiver operating characteristic curve ranging from 0.85 to 0.97.

Main Results:

• PTDMs successfully inferred chemical hazard profiles and identified potential target genes and diseases.
• Demonstrated predictive capabilities by identifying new potential targets for bisphenol A and aspirin.
• Validated predictions with molecular hypotheses and existing experimental evidence from published literature.

Conclusions:

• The developed PTDMs framework offers a powerful approach for chemical profiling in human health exposure assessments.
• This method shows significant potential for improving environmental hazard evaluations.
• The integration of biological networks enhances the accuracy and scope of toxicological assessments.