Figure 4

CloVR-ITS heatmap clustering visualizations for gastric fluid data. CloVR-ITS outputs hierarchical clusterings of samples and annotations at several taxonomic levels. These clusterings are computed utilizing the furthest-neighbor algorithm with a Euclidean-distance metric on either log-transformed proportions or simple proportions of taxon groups, shown at the class level in (A) and (B), respectively. While proportions alone help the user assess the overall distribution of taxonomic assignments for each sample, log-transformed proportions can be useful for visualizing similarities or differences between samples that affect members of the community that may make up less than 1% of the total sequences in a sample. Clusterings did not demonstrate significant associations among gastric fluid samples from subjects with similar treatments or conditions (see Additional file 2: Table S1 for sample metadata).