Analysis of Multiplexed Flow Cytometric Assays and Toxicogenomic Signatures for Genotoxicity Prediction: A Model...

Area of Science:

• Biomedical Science
• Toxicology
• Computational Biology

Background:

• Genotoxicity assays are crucial in drug development to assess cancer risk.
• Traditional assays offer limited mechanistic details, necessitating advanced methods.

Purpose of the Study:

• Evaluate machine learning (ML) models for genotoxicity and mechanism of action (MoA) prediction.
• Compare ML model performance using MultiFlow and MicroFlow assay data.
• Integrate toxicogenomic data to enhance understanding of genotoxicity mechanisms.

Main Methods:

• Implemented ML models with in-house MultiFlow and MicroFlow DNA damage assay data.
• Collected and analyzed dose-response data for improved assay inference.
• Utilized toxicogenomic data (TGx-DDI, RNA-seq) for case studies on complex compounds.

Main Results:

• ML models achieved high accuracy: 96% for MoA and 99% for genotoxicity prediction.
• Dose-response analysis improved assay inference and model precision.
• Toxicogenomics integration provided deeper mechanistic insights into genotoxicity pathways.

Conclusions:

• ML models integrated with MultiFlow and MicroFlow assays significantly improve genotoxicity prediction.
• Toxicogenomic data integration offers a powerful framework for detailed genotoxicity mechanism elucidation.
• This approach enhances precision in safety assessments during drug development.