Fig. 2

Classification performance of CAT for different levels of unknownness across a range of parameter settings. Thickness of markers indicates values of the f parameter; runs with similar r parameter values are connected with black lines. Markers indicate maximum and minimum values out of ten benchmarking datasets, bars cross at the means. Color coding indicates the mean taxonomic rank of classification averaged across the then benchmarking datasets (minimum and maximum values not shown for brevity). Gray lines in the plot depict sensitivity, which is defined as the fraction of classified sequences times precision. Runs with equal parameter settings are connected in the parameter settings figure, showing that CAT achieves a high precision regardless of unknownness of the query sequence, by classifying sequences that are more unknown at higher taxonomic ranks. Default parameter combination (r = 10, f = 0.5) is shown in red