In a recent investigation published in the journal Science, a synthetic gene library harboring silent mutations was employed to conclude that codon bias does not affect expression of the green fluorescent protein (GFP) in E. coli. This result is seemingly at odds with the more conventional thinking that use of codons influences the speed and accuracy of the elongation phase of translation. In their article, Supek and Dr. Šmuc argue that the conventional view is generally correct, substantiating their claims with a statistical re-analysis of the data from the GFP experiment using non-linear regression methods, decision trees and support vector regression.