Fig. 6.

Analyses of biologically-meaningful outputs are dependent on 16S rRNA sequence processing. a α diversity metrics vary greatly between OTU picking approaches, and are dependent on the choice of reference database in the case of reference-based OTU clustering methods. Black dotted lines indicate the expected values of these metrics based on known community composition. b-c Phylogeny-dependent β diversity metrics, including Weighted UniFrac (b) and Unweighted UniFrac (c), differ depending on the method of phylogeny-generation. A comparison of the distribution of samples via a Procrustes analysis indicates the impact that the phylogenetic tree makes on these data. d sl1p generates 2 phylogenies. The default phylogeny represents the phylogeny generated as part of the default QIIME workflow. The pruned phylogeny is generated by sl1p by pruning the Greengenes reference phylogeny to those branches which are present within the sample set