Establishing a BARCODE platform to categorize and disseminate active matter data

Active matter systems necessitate large multidimensional datasets to accurately represent the heterogeneous behaviors of various soft materials. Analyzing and extracting useful information from these large datasets in a tractable and meaningful manner is a major challenge in active matter research. Here, we discuss our further development of BARCODE (Biomaterial Activity Readouts to Categorize, Optimize, Design, and Engineer), a high throughput characterization tool for active matter microscopy videos. BARCODE pairs a modular design and established analysis techniques to extract a highly reduced set of key metrics for each video as well as information-rich Reduced Data Structures for detailed analysis. We build on BARCODE by introducing a number of new metrics to more robustly characterize structure and dynamics, providing low-dimensionality data fingerprints amenable to machine learning algorithms. We also develop a set of simulated experimental datasets with known characteristics to allow users to validate existing and new metrics they can add to BARCODE, and train their analysis workflows. Finally, we extend BARCODE capabilities to analyze simulation data, allowing for direct comparison of simulated and experimental results to provide more meaningful experimental-theory feedback loops.