Fig 1. Accuracy of identification and abundance for three binary mixtures of bacteria using Centrifuge, CLARK, and KrakenUniq.

After quality control, sequences were analyzed by Centrifuge, CLARK, and KrakenUniq using default settings and databases. Abundance estimates were filtered to include only organisms classified at the species or strain-level with at least 0.1% abundance. False positive abundance was calculated by summing the relative abundances of any organism identified by the classifiers that was not part of the mixture. The relative abundance of organisms identified by each tool is represented by circle size with actual values displayed below; values that are zero have no circle.