Analysis of oral bacterial communities: comparison of HOMI with a tree-based approach implemented in QIIME.

Molecular taxonomic assignments in oral microbial communities have been made using probe-matching approaches, but never compared to those obtained by more readily accepted tree-based approaches.  To compare community composition profiles obtained from a probe-matching approach (HOMI) to those from a closed-ended tree-based approach (QIIME using the eHOMD database).  HOMIand QIIME were used for parallel analysis of ten mock community samples, and of 119 supragingival plaque samples from ecologically unique sites (sound tooth surfaces in healthy subjects, sound tooth surfaces in patients with primary Sjögren's Syndrome, and carious lesions in Sjögren's Syndrome patients). Linear discriminant analysis Effective Size (LEfSe) was used to identify discriminating taxa among the natural plaque samples.  Community composition profiles of all samples were congruent between the two analysis aproaches. Alpha and beta diversity of the natural plaque communities were likewise similar. Communities from pSS patients and those from individuals with normal salivary flow differed in alpha and beta diversity. Both classification approaches yielded differences in composition predicted for samples from these subject cohorts, and discriminating taxa were similar between approaches.  A direct comparison demonstrates that HOMI is largely equivalent to the tree-based approach as implemented here.