Spatially Aware Plate Layouts for Optimal Plate Designs

In an upcoming talk, hosted by the American Statistical Association: Biopharmaceutical Section, Dr Beate Ehrhardt from the Institute for Mathematical Innovation, University of Bath, will describe how the spatially aware plate layout tool which she developed, can help improve reproducibility, reduce bias, and strengthen downstream statistical analyses.

This online talk is on Friday, 10 July at 16:00-17:00 BST and anyone with an interest is welcome to attend.

Title

Spatially Aware Plate Layouts (SAPL): An R Shiny Tool Integrating Experimental Constraints into Randomization for Optimal Plate Designs

Presenter

Dr Ehrhardt's research focusses on the statistical analysis and application of machine learning to interdisciplinary research in academia, industry and the third sector.

Abstract

Multi well plates underpin high throughput experimentation across biochemistry, pharmacology, and molecular biology, yet spatial artefacts-such as edge effects and local clustering-can introduce systematic biases that compromise reproducibility and statistical power. Traditional layout strategies, whether manually designed or based on naive randomization, often fail to address these spatial dependencies effectively.

SAPL is a controlled randomization framework that merges rigorous experimental design principles with the spatial structure of common plate formats. In contrast to full randomisation, which may inadvertently concentrate critical conditions or controls, SAPL actively enforces spatial balance by managing factors such as control dispersion, edge related susceptibility, and row/column structure. This approach enhances the reliability of normalisation procedures and strengthens downstream statistical analyses.

By integrating spatial awareness into the randomisation procedure, SAPL helps improve the reproducibility, robustness, and translational value of preclinical data. To facilitate widespread use, SAPL is available as an open access, user friendly R Shiny tool requiring no programming expertise. It reduces the time of generating a 384 well plate from 1 day to 15min. The framework has been rigorously evaluated in collaboration with scientists at AstraZeneca and the Functional Genomics Screening Laboratory across diverse experimental settings.