On the correct interpretation of similarity index in codon usage studies: Comparison with four other metrics and implications for Zika and West Nile virus

Viruses can modify the usage of their synonymous codons to mimic that of their hosts: investigation of codon usage may therefore help increase our understanding of virus biology. The similarity index (SiD or (D(A,B)) is a recent method for estimating the resemblance of the overall codon usage of the hosts to that of a certain virus. Here we report that this index has been systematically misinterpreted in the literature. We first review its mathematical meaning to show that a high value of SiD indicates low codon usage similarity, and not the opposite as frequently reported. We further compared SiD with four other metrics - nCAI, nRCDI, I_TE and tAI - using Zika and West Nile arboviruses in relation to various vectors and hosts. We highlight some differences between the five metrics and present a first bioinformatic evidence that Zika virus is poorly adapted to Culex quinquefasciatus, in accordance with the known biology of this virus. We conclude that SiD, if correctly interpreted, is a useful tool for codon studies, but we advocate that the results from this and other methods are context and time dependent and should be taken with care. A script to automatically calculate SiD from two codon usage tables is available at https://github.com/andrea-silverj/SiDcal.