Skip to main content
. 2011 Jun 28;6(6):e21105. doi: 10.1371/journal.pone.0021105

Figure 7. Predictions in the human-gut urn when simulating the rare biosphere.

Figure 7

In a sample of size Inline graphic from a human-gut, Inline graphic species were discovered. Based on our methods, we estimate that Inline graphic of these species represent Inline graphic of that gut environment; hence, the remaining Inline graphic is composed by at least Inline graphic species. To test our predictions of the conditional uncovered probability (black), we simulated the rare biosphere by adding additional species and hypothesized that our point prediction could be offset by up to one order of magnitude: point predictions produced by the Embedding Algorithm (blue), point predictions produced by the algorithm each time a new species was discovered (red), Inline graphic upper-bound (orange), and Inline graphic conservative-upper interval (green). The predictions used the parameters Inline graphic. The different urns were devised as follows. For each Inline graphic (indexing rows) and Inline graphic (indexing columns), a mixture of two urns was considered: an urn with the same distribution as the microbes found in the gut dataset, and weighted by the factor Inline graphic, and an urn consisting of Inline graphic colors (disjoint from the gut urn), with an exponentially decaying rank curve and weighted by the factor Inline graphic. See Fig. 5 for the rank curve associated with each urn.