FIGURE 1. Growth phenotype study.

a, Comparison of high-throughput phenotyping array data (Exp) with predictions for the E. coli network, both considering regulatory constraints (Reg) and ignoring such constraints as a control (Met). Each case is categorized by comparison type (Exp/Met/Reg), and results are listed as ' + ' (predicted or observed growth), ' - ' (no growth) or 'n' (for cases involving a regulatory gene knockout not predictable by the Met model). The comparisons are further divided into four subgroups represented by different colours. b, Chart showing individual results for each knockout under each environmental condition, with results categorized and coloured as in a. The environments involve variation of a carbon or nitrogen source and are further divided into subgroups: AA, amino acid or derivative; CM, central metabolic intermediate; NU, nucleotide or nucleoside; SU, sugar; OT, other. The knockout strains are also divided by functional group: A, amino acid biosynthesis and metabolism; B, biosynthesis of cofactors, prosthetic groups and carriers; C, carbon compound catabolism; P, cell processes (including adaptation and protection); S, cell structure; M, central intermediary metabolism, E; energy metabolism; F, fatty acid and phospholipid metabolism; N, nucleotide biosynthesis and metabolism; R, regulatory function; T, transport and binding proteins; U, unassigned. Each environment and knockout strain is associated with a fraction of agreement (FA) between regulatory model predictions and observed phenotypes, as shown in the bar charts to the right and below. c, Table showing all environments or knockout strains for which FA < 0.60. Of these substrates or knockout strains, 18 point to uncharacterized metabolic or regulatory capabilities in this organism, as indicated (see Supplementary Information for a case-by-case analysis).