Hypothesis testing in high-throughput screening for drug discovery.

Area of Science:

• Biochemistry
• Bioinformatics
• Drug Discovery

Background:

• High-throughput screening (HTS) is revolutionizing biological sciences, but analyzing its vast data requires advanced statistical tools.
• Current HTS methods often characterize hit lists retrospectively, limiting real-time quality control and resource optimization.
• Statistical tools for DNA chip studies are powerful but underutilized in compound screening.

Purpose of the Study:

• To develop a method for predicting false positive and false negative rates directly from primary HTS results.
• To enhance prioritization, resource utilization, and quality control in HTS campaigns.
• To adapt p-value distribution analysis (PVDA) from gene expression studies for HTS data.

Main Methods:

• Applied p-value distribution analysis (PVDA), originally for gene expression studies, to HTS data.
• Utilized information from compounds and controls in HTS results and replicated pilot runs.
• Estimated Z scores and p-values for each measurement within the HTS data.

Main Results:

• PVDA enabled the prediction of all relevant error rates, including false positives and false negatives.
• The method also allowed for the prediction of the rate of true inactives.
• PVDA predictions showed excellent agreement with subsequent confirmation experiments.

Conclusions:

• PVDA is a valuable tool for real-time analysis and quality control in HTS.
• This approach optimizes resource allocation and improves the reliability of HTS results.
• PVDA offers a robust method for predicting error rates and true inactives in compound screening.