Clustering of the 176 depth-related COGs.

(A) Hierarchical clustering of the 176 depth-related COGs in the 24 datasets. Clustering analysis is based on the normalized abundance profile of the 176 depth-related COGs that were shared by the three reference water columns (ATII, ALOHA and BATS) and significantly differed in abundance within at least one of them (details in Materials and Methods). The height indicates the relative distance between datasets. Bootstrap confidence values above 60 for the nodes are shown. The heatmap is shown in fig. S3A. (B) Location of the boundary between the photic and aphotic zones in each of the four water columns. The arrows indicate the depth at which PAR reaches 1% of the level at the surface. (C) Hierarchical clustering of the photic/aphotic global-core depth-related COGs. 82 COGs that showed statistically significant difference (Welch's test, FDR-corrected, p≤1E-04) in their normalized abundance between the photic and the aphotic groups of datasets were selected to establish a photic/aphotic global-core, depth-related reference set. 54 COGs had significantly higher abundance in the photic datasets (Group I, table S4A in file S1); contrary to the remaining 28 aphotic related COGs (Group II, table S4B in file S1). Bootstrap confidence values for the major nodes are shown. Heatmap coloring reflects the Z score of normalized abundances of each COG across all clustered datasets (details in Materials and Methods).