Open-source low-cost liquid handlers for the development of self driving laboratories

The use of rational trial and error experimentation becomes inefficient as desired experimental design becomes increasingly complex. Often, desirable traits are dependent on multiple variables requiring simultaneous optimization for single and multi-objective targets. The experimental design described thus becomes a time intensive and expensive endeavor where a multitude of dynamic interactions are not captured when obtaining the resulting functional outcomes. Modern data science tools such as machine learning and active learning supplemented by robotics provide an invaluable opportunity to alleviate these experimental inefficiencies. However, currently there is a lack of accessible automation tools for the purpose of widespread implementation of self-driving workflows. To meet this need for democratization of relevant technologies, open-source, low-cost liquid handling platforms were created to offer practical templates for researchers adopting self-driving lab methodologies. This device was validated through the implementation of a campaign of automated experimentation and data analysis to optimize a representative enzyme activity assay. The resulting data from the developed and validated self-driving laboratory plays an important role in providing simple but non-trivial demonstration of the ability of automated workflows to efficiently predict the influences of input parameters on output properties of interest.