Reply to Sun et al., “Identifying Composition Novelty in Microbiome Studies: Improvement of Prediction Accuracy”

To quantitatively measure the beta diversities between microbiomes, Microbiome Search Engine (MSE) (1) calculates phylogeny similarity using operational taxonomy unit (OTU) profiles; for both query and database samples, all 16S rRNA gene sequences are mapped to the Greengenes database (version 13-8) (2) for reference-based OTU picking with a 97% cutoff. Thus, in MSE, the comparison between query and database samples is approximately at the species level (3), although the actual taxonomic resolution varies according to taxon, due to differences in the evolutionary rates of the 16S rRNAs. Moreover, in MSE, both the relative abundance (with 16S rRNA gene copy number normalization [4]) and the phylogenetic structures of OTUs are utilized for similarity calculation (as in UniFrac [5, 6]), yet the speed is optimized by nonrecursive computing to enable real-time responses (7). By comparing the query sample (i.e., dust from university dormitories) provided by Sun …