Theoretical models for heterogeneity of base composition in DNA

Data on DNA base composition heterogeneity and molecular size, mainly from bacteria, are compiled from the literature. A series of theoretical models of increasing complexity is put forward, and it is concluded that the data support a model in which the DNA consists of a sequence of “segments” with different underlying base compositions. Estimates of mean segment length and variance of G + C content are made. The possibility that the excess heterogeneity above random results from variation in the composition of coded proteins is considered and rejected. It is concluded that the heterogeneity is probably caused by variations in the relative use of synonymous codons in different genes.