Fig. 5

An illustration of replacing low abundance OTU with 0 value, Rand index plot, PCoA plots, and the total Shannon diversity plot of the modified Smits dataset. A A schematic plot illustrates how we replace low abundance values with a 0 value when the threshold is set to be 1. B Clustering performance of unweighted UniFrac improves by increasing the threshold used to change non-zero entries to 0 in the Smits dataset, while the performance of Bray Curtis remains the same. C Unweighted UniFrac beta diversity PCoA of the original data, the dataset where entries less than 30 are converted to 0, and the dataset where entries less than 60 are converted to 0. D The total Shannon diversity of versions of the Smits dataset decreases as we raise the threshold, below which counts are converted to 0