Computer-aided Biology: DoE Solutions

Assay development is a multi-parametric problem, with non-linear interplay between interacting factors. For biochemical assays, many factors are determined by physical constants, such as the K_m of a substrate or k_cat of the reaction. However, the choice of the reaction buffer is more complex. Here we share our development of an automated platform to specifically select for the best performing buffers in biochemical assays, applicable to all types of enzymes. Design of experiments (DoE), a statistical methodology, was used in combination with Dragonfly robotics by SPT Labtech.

Using DoE, results were delivered in hours instead of months. Multiple combinations of variables were tested, which would otherwise be omitted to save time in traditional one-factor-at-a-time experiments. In our buffer screens, we found that the signal windows of the reactions could be enhanced drastically, to more than 100 times the original starting buffer conditions. Furthermore, we demonstrated meaningful savings in the use of reagents, the time taken to run the assays, and an increase in the robustness of assays. In this talk, case studies of assay improvements will be shared, including methyl transferase, HTRF kinase, molecular glue and exonuclease assay.