Improved Statistical Methods for Hit Selection in High-Throughput Screening

Abstract

High-throughput screening (HTS) plays a central role in modern drug discovery, allowing the rapid screening of large compound collections against a variety of putative drug targets. HTS is an industrial-scale process, relying on sophisticated automation, control, and state-of-the art detection technologies to organize, test, and measure hundreds of thousands to millions of compounds in nano- to microliter volumes. Despite this high technology, hit selection for HTS is still typically done using simple data analysis and basic statistical methods. The authors discuss in this article some shortcomings of these methods and present alternatives based on modern methods of statistical data analysis. Most important, they describe and show numerous real examples from the biologist-friendly Stat Server HTS application (SHS), a custom-developed software tool built on the commercially available S-PLUS and StatServer statistical analysis and server software. This system remotely processes HTS data using powerful and sophisticated statistical methodology but insulates users from the technical details by outputting results in a variety of readily interpretable graphs and tables.

Statistical and Graphical Methods for Quality Control Determination of High-Throughput Screening Data

“…The details of this calculation are described more fully in the appendix to the companion article. 1 There was clearly a major shift in the median of the raw values around plate 100 from about 4000 to about 9000 or 10,000. Note also that this shift occurred when switching from one runset to the next.…”

“…Adjusted values for each plate are computed by subtracting the systematic positional effects calculated in step 1 from the raw data (note that controls are not used in these calculations at all, as they can be distorted in unknown ways by positional effects, themselves). The algorithm then further adjusts the adjusted values for each well longitudinally, that is, over the plate sequence, using the robust smoothing procedure described in Brideau et al 1 and illustrated by the overlaid trend curves of Figures 2 to 4. This allows for certain kinds of local positional effects that cannot be adequately determined in step 1.…”

“…Hence, the variability seen among them is mostly just the intrinsic variability in the impotent compounds plus short-term assay variability. A resistant estimate, the "mad" (median absolute deviation from the median 1,3 ) is used to estimate this variability. A resistant estimate is used so that a few outliers that might well be potent compounds do not inflate the estimate.…”

From experimental setup to bioinformatics: An RNAi screening platform to identify host factors involved in HIV‐1 replication

“…This is done for each plate individually, since effects may be different from plate to plate. Spatial effects within each plate are then corrected using B-score normalisation [18]. This method adjusts intensities using a median polish procedure on rows and columns separately, thus estimating row-and column-effects on each plate.…”

Functional genomic analysis of glioblastoma multiforme through short interfering RNA screening: a paradigm for therapeutic development

“…9 Screening typically relies on sophisticated automation, appropriate controls, and state-of-the-art detection technologies to organize and analyze thousands of test samples. 5,7 Unfortunately, siRNA screening is associated with "off-target" effects, which can affect the analysis and final results. Data output, therefore, requires sophisticated and rigorous analysis methods to reduce the number of high-confidence "hits."…”