Fig. 1

Spearman correlations between metagenome predictions and shotgun metagenome sequencing in unpermuted and permuted datasets. a Each gene’s abundance was permuted across samples independently. b and c An example showing the correlations between genes relative abundances estimated by PICRUSt and metagenome sequencing in a soil sample (sample BulkAG3 in soil_AAN dataset) for unpermuted (b) and permuted (c) data. d–f The Spearman correlations of gene composition estimated from metagenome sequencing and predicted with PICRUSt (d), PICRUSt2 (e), and Tax4Fun (f) in unpermuted (red) and permuted data (blue) in all datasets. In each of the 100 permutations, every gene’s abundance was permuted across samples independently