Quality Control of Quantitative High Throughput Screening Data.

Area of Science:

• Pharmacology
• Toxicology
• Computational Biology

Background:

• Quantitative high throughput screening (qHTS) generates extensive concentration-response data for compound profiling.
• Potency estimates (AC50) can be unreliable due to variable response patterns for individual compounds.
• Need for robust quality control in qHTS to ensure data accuracy and reliable compound assessment.

Purpose of the Study:

• Introduce an automated quality control method, CASANOVA (Cluster Analysis by Subgroups using ANOVA).
• Improve potency estimation by filtering compounds with inconsistent concentration-response patterns.
• Enhance the reliability of adverse effect screening in qHTS.

Main Methods:

• Developed CASANOVA, a procedure utilizing analysis of variance (ANOVA) for response pattern clustering.
• Applied CASANOVA to 43 public qHTS datasets to group compound-specific responses.
• Conducted extensive simulation studies to evaluate error rates and AC50 estimation accuracy.

Main Results:

• Only ~20% of compounds exhibited single-cluster responses when analyzed with CASANOVA.
• CASANOVA demonstrated low error rates (<5%) for cluster separation and merging.
• Weighted averaging of AC50 estimates yielded results within a 10-fold range.

Conclusions:

• CASANOVA effectively identifies and filters compounds with inconsistent response patterns in qHTS.
• The method significantly improves the trustworthiness of potency (AC50) estimations.
• CASANOVA enhances the overall quality and reliability of qHTS data analysis for drug discovery and toxicology.