Figure 1.
Read classification at the species, genus, and family levels for all three long-read data sets. Precision and recall are shown for all six compared tools on three taxonomic levels. Taxor and KMCP show the highest precisions across all three data sets and all levels. Taxor's recall is comparable to MetaMaps and Ganon on the species level, whereas KMCP consistently has the lowest recall of all tools. Centrifuge and Kraken2 have low species-level recall for the two ONT data sets while performing well on the HiFi data set.
