ANALYSIS

Data from individual HOMIM signals are translated to a “bar code” format and normalized by comparing individual signal intensities to the average of signals from universal probes (Fig. 1).

Figure 1: Bacterial profile map or "bar code" of clinical samples from a single mouth. (A) Bacterial profiles of over 200 bacterial species comparing 10 samples from a single mouth. (B) Magnified from (A) illustrating dendrogram of cluster analysis and partial list of bacterial species tested. Band intensities represent relative bacterial proportions.

The bands correspond to presence or absence and band intensities are scored as 0, 1+, 2+, 3+, 4+ or 5+. These data are provided in Excel tables. For details, see "Data Analysis Procedure of the HOMIM online tool." Thus, more intense bands reflect higher proportions of a given species. Differences or similarities in profiles can readily be seen. Samples are subsequently subjected to cluster analysis (Dr. George Chen). The dendrogram of the cluster analysis (top of figure) is shown more clearly in Figure 1B. The cluster program sorts the profiles into the appropriate clusters. It is noteworthy in this analysis of sites from a single mouth that samples 4 and 5, which clustered tightly together, were samples from the left side of the tongue and the right side of the tongue, illustrating the level of reproducibility with the assay.  These data are what you would expect to see.

Consequently, bacterial profiles of 100's of HOMIMs of clinical samples can be readily analyzed as shown in Figure 2.

Figure 2: Bacterial profiles of 461 bacterial taxa (representing about 300 species) comparing subgingival plaque from 105 healthy sites in periodontally healthy subjects (n=20) to 154 diseased sites from periodontally diseased subjects (n=47). Differences in profiles can be seen at a glance. (Courtesy of AP Colombo)

A final report and data analyses are emailed to each user. For registered users, data can be also accessed online through a dedicated website. In addition, data are cumulative, so analyses of old data can be potentially compared with new data.