SUMMARY Analysis of compositions of microbiomes (ANCOM) compares the absolute abundances of microbes between two or more ecosystems using relative abundances in specimens derived from these ecosystems. Despite its impressive… Click to show full abstract
SUMMARY Analysis of compositions of microbiomes (ANCOM) compares the absolute abundances of microbes between two or more ecosystems using relative abundances in specimens derived from these ecosystems. Despite its impressive performance, there are two drawbacks to ANCOM. First, with K microbes it requires fitting K(K-1)/2 models for log-ratios of counts, and so can be computationally intensive. Second, it does not output p-values for microbes detected as differentially abundant. We propose a fast implementation of ANCOM, fastANCOM, that fits only K models for log transformed counts. fastANCOM provides p-values to declare statistical significance and outputs log fold changes of abundance between groups. We demonstrate that fastANCOM compares favorably with existing differential abundance testing methods in terms of running time, false discovery rate, and power. AVAILABILITY AND IMPLEMENTATION fastANCOM is available at https://github.com/ZRChao/fastANCOM. SUPPLEMENTARY INFORMATION Supplementary data are available at Bioinformatics online.
               
Click one of the above tabs to view related content.