PROJECT SUMMARY The microbiome plays an important role in many human disorders and diseases, including cancer, autoimmune diseases and sexually transmitted infections. Microbial communities are usually studied via their genetic sequences, and a major challenge in modern microbiome science is that microbial sequencing introduces both bias and variability. The magnitude of sample- and study- specific variation in sample measurements can exceed the magnitude of variation due to treatment or disease status, which impedes the diagnosis and treatment of complex diseases. To support low-cost, rigorous and reproducible microbiome research, we will develop statistical tools to guide researchers in decision-making in the presence of measurement error in microbiome studies. We will focus specifically on robust approaches to differential abundance, integrated models for multiple biological units (including both within- and across-kingdom interactions), and methods for emerging data structures (such as samples with spike-in cells or communities). We will also develop recommendations on the experimental design of microbiome studies, focusing on maximizing statistical power to make true discoveries while minimizing sequencing and labor costs. Our methods apply to a broad range of microbial questions, including ecology, metabolism, evolution, and community assembly. Our methods will be accompanied by freely available, open-source software, as well as detailed tutorials and forums for user questions. The long-term goal of this research is to improve the efficiency with which microbiome science can have a positive impact on the etiology and treatment of human disease and infection.