Figure 6

Predictive values for E and S plotted against amino acid hydrophobicity. The abscissa gives the hydrophobicity of each amino acid on the Kyte-Doolittle scale. The ordinate gives the predictive values for the fractional content of each amino acid as calculated from single logistic regressions performed against E or S values in the Analysis Dataset. (See the legend to Figure 4 for a definition of the predictive value.) The error bars show 95% confidence. Predictive values for total amino fractions are shown as solid squares. For the four amino acids commonly considered to be encoded in part by rare codons, the predictive values are also shown for the corresponding amino acid fractions encoded by either common codons (triangles pointing up) or rare codons (triangles pointing down); see Additional file 1, Figure S4 and the associated text in the Results section for an explanation of this segregation procedure. Amino acid hydrophobicity is not significantly correlated with amino acid predictive value for expression (p = 0.098) or solubility (p = 0.23).