Predicting Prokaryotic Ecological Niches Using Genome Sequence Analysis
Figure 3
Percentage of comparisons with matches for each prokaryote based on a shared taxonomic group metric.
The top 5, 10, 15, 20, and 25 nearest neighbors for each prokaryotic species were retained and their taxonomic groups compared by computing the average number of matches. This analysis was performed for prokaryotes on the niche map (circles) and the phylogenetic map (squares) as shown. In all cases, sequencing bias was taken into account by disregarding all pairs of prokaryotic species that had a Jaccard coefficient greater than 0.90 based on Pfam profile analysis (see Materials and Methods). A control (average of 10 trials) was also performed by randomizing the nearest neighbors for each prokaryotic species (hashes). Error bars for the randomized control are too small to be displayed.