ProgPerm: Progressive permutation for a dynamic representation of the robustness of microbiome discoveries

Identification of significant features is a critical task in microbiome studies that is complicated by the fact that microbial data are high dimensional and heterogeneous. Masked by the complexity of the data, the problem of separating signal from noise becomes challenging and troublesome. For instance, when performing differential abundance tests, multiple testing adjustments tend to be overconservative, as the probability of a type I error (false positive) increases dramatically with the large numbers of hypotheses. We represent the significance identification problem as a dynamic process of separating signals from a randomized background. The signals and noises in this process will converge from fully mixing to clearly separating, if the original data is differential by the grouping factor. We propose the progressive permutation method to achieve this process and show the converging trend. The proposed method progressively permutes the grouping factor labels of microbiome and performs multiple differential abundance tests in each scenario. We compare the signal strength of top hits from the original data with their performance in permutations, and will observe an apparent decreasing trend if these top hits are true positives identified from the data. To help understand the robustness of the discoveries and identify best hits, we develop a user-friendly and efficient RShiny tool. Simulations and applications on real data show that the proposed method can evaluate the overall association between microbiome and the grouping factor, rank the robustness of the discovered microbes, and list the discoveries, their effect sizes, and individual abundances.