Round Trip: An Automated Pipeline for Experimental Design, Execution, and Analysis

Synthetic biology is a complex discipline that involves creating detailed, purpose-built designs from genetic parts. This process is often phrased as a Design-Build-Test-Learn loop, where iterative design improvements can be made, implemented, measured, and analyzed. Automation can potentially improve both the end-to-end duration of the process and the utility of data produced by the process. One of the most important considerations for the development of effective automation and quality data is a rigorous description of implicit knowledge encoded as a formal knowledge representation. The development of knowledge representation for the process poses a number of challenges, including developing effective human-machine interfaces, protecting against and repairing user error, providing flexibility for terminological mismatches, and supporting extensibility to new experimental types. We address these challenges with the DARPA SD2 Round Trip software architecture. The Round Trip is an open architecture that automates many of the key steps in the Test and Learn phases of a Design-Build-Test-Learn loop for high-throughput laboratory science. The primary contribution of the Round Trip is to assist with and otherwise automate metadata creation, curation, standardization, and linkage with experimental data. The Round Trip's focus on metadata supports fast, automated, and replicable analysis of experiments as well as experimental situational awareness and experimental interpretability. We highlight the major software components and data representations that enable the Round Trip to speed up the design and analysis of experiments by 2 orders of magnitude over prior