Results of VI-cut compared to standard methodologies. (a) Standard methodologies using a specific MSA with furthest neighbor clustering to find OTUs. VI-cut was employed using the same MSA and distance correction in each plot. For each VI-cut trial, 10% of the sequences were randomly selected and given labels. Over 20 trials, OTUs determined by VI-cut are stable and more accurate than the standard methodologies. (b) Distribution of distances within clusters defined by the species-annotation and a sample VI-cut clustering. Clusters with one member (i.e. singletons) are not shown. There is considerable variation (D = 0.01-0.07) in the optimal distance threshold among species.