Optimizing androgen receptor prioritization using high-throughput assay-based activity models.

Area of Science:

• Environmental toxicology
• Computational toxicology
• Chemical screening

Background:

• Computational models aid chemical prioritization for the U.S. EPA's Endocrine Disruptor Screening Program (EDSP).
• Existing androgen receptor (AR) pathway models require optimization for efficiency and cost-effectiveness.
• Assay availability and chemical diversity pose challenges for current screening models.

Purpose of the Study:

• To demonstrate a data processing method for determining optimal minimal assay batteries.
• To establish a uniform evaluation method for minimal assay batteries against AR pathway models.
• To integrate chemical cluster analysis into assay battery performance evaluation.

Main Methods:

• Compared two previously published AR pathway models (11- and 14-assay).
• Investigated assay subsets to optimize testing strategies for cost and sensitivity.
• Incorporated chemical structure-based clustering into a multi-stage testing workflow.

Main Results:

• An expanded 14-assay model showed higher sensitivity for antagonists; the 11-assay model favored agonists.
• Identified critical assays: 3 for antagonism, 2 for agonism.
• A minimum of 9 assays are needed for 95% sensitivity for both agonism and antagonism.
• Chemical clustering reduced the average assays needed per chemical by 52% in a multi-stage workflow.
• In silico predictions further reduced resource requirements.

Conclusions:

• A data-driven approach using chemical clustering and multi-mechanism consideration enhances chemical screening efficiency.
• This case study validates a proof-of-concept for optimizing assay batteries under the EDSP.
• Efficient screening maximizes chemical throughput and enables data-driven prioritization for further testing.