Figure 4. Classification results for the 60 bp read set.

(A) Mean sensitivity versus mean precision. The mean sensitivities (Sensitivity_s & Sensitivity_s&h) are the means of the proportions of reads correctly classified over the total number of simulated reads across viruses. The mean precisions (Precision_s & Precision_s&h) are the means of the proportions of reads correctly classified over the number of classified reads across viruses. Circles denote the values if only “correct species” reads are considered as correctly classified reads; triangles denote the values if “correct species” and “correct higher” reads are considered as correctly classified reads (see Materials and Methods). The perfect classifier would have 100% sensitivity and 100% precision. (B) Total number of viruses recovered for each classifier when correctly identifying at least 1 read per virus. The dashed line indicates the total number of tested viruses (233). (C) Mean number of spurious extra taxa per classifier. In this plot, a taxon is assumed as identified by a classifier if at least 1 read is assigned to it.