GOLDBAR: A Framework\nfor Combinatorial Biological\nDesign

With the rise of new DNA part libraries and technologies\nfor assembling\nDNA, synthetic biologists are increasingly constructing and screening\ncombinatorial libraries to optimize their biological designs. As combinatorial\nlibraries are used to generate data on design performance, new rules\nfor composing biological designs will emerge. Most formal frameworks\nfor combinatorial design, however, do not yet support formal comparison\nof design composition, which is needed to facilitate automated analysis\nand machine learning in massive biological design spaces. To address\nthis need, we introduce a combinatorial design framework called GOLDBAR.\nCompared with existing frameworks, GOLDBAR enables synthetic biologists\nto intersect and merge the rules for entire classes of biological\ndesigns to extract common design motifs and infer new ones. Here,\nwe demonstrate the application of GOLDBAR to refine/validate design\nspaces for TetR-homologue transcriptional logic circuits, verify the\nassembly of a partial <i>nif</i> gene cluster, and infer\nnovel gene clusters for the biosynthesis of rebeccamycin. We also\ndiscuss how GOLDBAR could be used to facilitate grammar-based machine\nlearning in synthetic biology.