Fig. 2

Analysis of the classified reads by Kraken, Kraken 2 and MEM approaches (Kaiju and PRROMenade). a Stacked bar chart to show the percentage of reads that were classified and non-classified using the complete IFGP database by Kraken, Kraken 2 and MEM approaches using simulated reads with 5, 10 and 20% substitution rates. The classified reads are divided into categories to show the proportions of the classified reads that were assigned to either the root, an incorrect GO term or finally to a correct or related GO term that was not the root LCF. b Stacked bar chart to show the percentage of reads that were classified and unclassified using the complete IFGP database by Kraken, Kraken 2 and MEM approaches using four experimental samples AV110 and AV112 (moist tropical forest) and S234 and S238 (boreal forest). The classified reads are divided into categories to show the proportions of the classified reads that were assigned to a GO term or the root