After data processing, potential therapeutic candidates are identified and categorized as strains, metabolites, or peptides and proteins. While administration of health-associated bacterial strains could potentially be used as therapies, Second Genome focuses on bioactive products derived from these strains that will have positive disease-modifying effects. These metabolites, peptides and proteins can be tested for activity in traditional pharmaceutical in vitro assays and further developed using classical drug development processes. Results of these in vitro assays are analyzed using machine learning models to improve subsequent nominations of therapeutic candidates.