Figure 2.

MIRAGE's predictive performance on reconstructing a known metabolic network of E. coli. The precision and recall of MIRAGE is marked with a star symbol. The precision and recall of several controls, including variants of MIRAGE that utilize only phylogenetic data, only expression data, or no functional-genomics data, are marked with a triangle, bar, and circle, respectively. The predictive performance of the functional genomic data (that is, by ordering potential gap-filling reactions based on their computed functional genomic weights, without utilizing metabolic flux analysis) is shown by the straight lines: the performance of the phylogenetic data, gene expression, and both data sources are colored green, yellow, and purple, respectively. The performance of random predictions of gap-filling reactions is colored blue.