Figure 2

Accuracy of new methods for the full-length case. 2a) In the ideal case (full-length queries with true alignment and no artificially introduced noise) Phyclass classified 66.4% of them correctly, with a standard deviation of 1.7%. For comparison, we show (second bar) the number of taxa placed correctly under Minimum Evolution phylogenetic analysis of the data set (71.1% correct, with a standard deviation of 1.6%). The third bar shows the percentage correct (52.2%, with standard deviation of 1.4%) when Phyclass was run on the same sequence set aligned by HMMER. 2b) Shows the distribution of node distances for classifications in the ideal case (Node Distance is defined as the number of nodes in the phylogenetic tree intervening between the correct placement and the inferred placement.) We see that 47% of incorrect estimates are off by only one node in the tree . Error bars represent ± 1 standard deviation of individual estimates, based on ten replicates.