Abstract
High throughput screening (HTS) campaigns often rely on a single test to make a decision to proceed onto the next step. Minimizing assay variability has thus become essential to obtain high quality data in order to improve decision making. Of the common sources of variability (e.g., liquid handling, detection, innate biological variability, random variability), liquid handling is often underappreciated, yet can be minimized by employing the right tools and strategy. Many laboratories perform liquid handling quality control (QC) by addressing each piece of equipment individually. While individual equipment QC certainly helps, we have found by considering the entire system in an approach known as process optimization, further gains in data quality can be realized. Process optimization assumes the entire assay assembly system, and not just individual components, contributes to variability. In other words, process optimization considers the order of liquid addition, pairing of reagents to liquid handlers, mixing, incubations, plate movement, etc.
This study was designed to examine the effect of liquid handling variability on a model biological assay by comparing non-optimized liquid handlers, optimized liquid handlers, and process optimization. The biological assay was composed of a binding protein (streptavidin) and a labeled ligand (biotin-fluorescein). Upon binding of biotin-fluorescein to streptavidin, the fluorescence signal decreases. Conversely, competition with an inhibitor results in an increased fluorescence signal. Three liquid handlers (e.g., manual pipetting, bulk peristaltic dispenser, and automated pipetter) were chosen to represent a typical HTS system in which each reagent was aspirated and dispensed by a different liquid handler. The first phase included randomly assigning a dispense method to a reagent. For example, a manual pipetter dispensed the streptavidin, the bulk dispenser dispensed the biotin-fluorescein reagent, and the automated pipettor dispensed the inhibitor. All reagent dispenses were 25 µL each to minimize volume bias. The second phase included optimizing each liquid handler individually by employing the Artel MVS (Multichannel Verification System). The assay was then performed using the same liquid handlers dispensing the same reagents. Finally, during the third phase, the liquid handlers were studied as part of a system by employing a design of experiment approach. The liquid handlers were then assigned the appropriate reagent to dispense. During all three phases, the assay was run and assay performance statistics were calculated and compared. This poster presents two major findings. First, how to study liquid handlers as part of a system within the context of the assay being optimized. Second, it was possible to further optimize the overall performance of the assay by not only optimizing the liquid handlers, but by choosing the correct liquid handler-reagent pairing. By studying the entire system, we were able to determine appropriate mixing and incubation parameters
![](/wp-content/uploads/2019/10/kathleen-chase.png")
