Figure 2

Hypergeometric probability of KEGG pathways among high confidence predictions. To determine which KEGG pathways have significant ratio of correct predictions among the highest confidence predictions, a hypergeometric probability analysis was performed. Each point on the graph represents a different KEGG pathway. On the x-axis is the hypergeometric probability that the pathway has more correct predictions than would be expected by chance, given the total number of predictions involving a member of the pathway. On the y-axis is the percentage of correct functional linkages involving members of the given pathway. By comparing the performance of profiles containing only prokaryotic organisms to profiles which also contain eukaryotes, it is evident that: (A) when E. coli was used as the query organism against Prok180 and Mix203, there is very little difference in performance. (B) When S. cerevisiae was used as query organism against Prok180 and Mix203, a large decrease in functional diversity, as measured by the number of significantly enriched KEGG pathways, was observed in the Mix203 set. Furthermore, there is also a marked decrease in accuracy even among those pathways which remain significantly enriched using the Mix203 set. (C) When S. cerevisiae was used as query organism against Euk23, while a small number of pathways were significantly enriched, they all had very poor accuracy, which greatly decreases the usefulness of their predictions as the false positives greatly outnumber the true positives.