A method to evaluate the relative weights of structural and functional constraints on a genome

A method that numerically evaluates the relative influence of structural or functional constraints on the codon composition of a gene is presented. The method applies to any constraint provided that it can be translated into a priori probability table of codon usage. The method is very easy to apply to different constraints and to different pools of sequences. Theoretical codon usage obtained from a linear combination of the various constraints was compared with experimental codon frequencies of several classes of genes taxonomically homogeneous and, applying a minimization tecnique, the relative weights of the examined constraints have been evaluated. The results show that a common strategy exists for all coding sequences even though most of the constraints are represented by different probability tables for different pools of genes. We also demonstrate that the method is reliable when it is applied to a single gene, so that it can also be used as a tool to predict the characteristics of a gene from its sequence. © 1991 Società Italiana di Fisica.