Introducing Bayesian thinking to high-throughput screening for false-negative rate estimation.

Area of Science:

• Drug Discovery
• Chemical Biology
• Bioinformatics

Background:

• High-throughput screening (HTS) identifies biologically active compounds but often lacks replication due to cost.
• Unreplicated primary screens risk losing true active compounds as false-negatives due to experimental variability.
• Accurate estimation of false-negatives is crucial for assessing screening quality and optimizing hit confirmation strategies.

Purpose of the Study:

• To develop and validate a computational method for predicting false-negative rates in HTS campaigns.
• To estimate the number of true active compounds and potential false-negatives from primary screening data.
• To provide a valuable tool for guiding compound selection and improving the efficiency of hit confirmation.

Main Methods:

• Implementation of a pilot screen (1% of library) to gather assay variability and preliminary activity data.
• Development of a predictive algorithm integrating Bayesian logic and Monte Carlo simulation.
• Application of the developed strategy to five distinct screening projects.

Main Results:

• The pilot screen provided essential data on assay variability and hit distribution.
• The Bayesian and Monte Carlo simulation-based algorithm successfully estimated false-negative rates.
• The method demonstrated utility in predicting the number of missed hits across multiple screening projects.

Conclusions:

• The developed computational strategy offers a reliable approach to estimate false-negative rates in HTS.
• This method enhances the understanding of screening quality and aids in optimizing resource allocation for hit confirmation.
• Predicting false-negatives is essential for maximizing the success of drug discovery screening campaigns.