All Times EDT
Keywords: Microbiome, 16S rRNA Sequencing, Association Analysis
In this presentation, I will discuss my methodology work in developing a powerful association test to determine if microbial community is related to clinical outcome or diseases. Most microbial association test methods can be categorized into two groups – OTU-level tests and community-level tests. The former focuses on testing and detecting single OTUs that are associated with diseases. The latter aims at determining if the whole microbial community is related to the clinical outcome. Although a lot of effort has been put in each group of methods, little literature has been seen to unite them. In this talk, I propose a new test framework that adaptively combines OTU-level tests into a community-level test. As a proof of concept, we combine score tests and compare its performance with MiRKAT and aMiSPU using simulation and real data analysis. We show that our method enjoys comparable or even better statistical power than methods specifically designed for community-level test.