Profiles of sequence compositional complexity provide a view of the spatial
heterogeneity of symbolic sequences at different levels of derail. Sequenc
e compositional complexity profiles are here decomposed into partial profil
es using the branching property of the Shannon entropy. This decomposition
shows the complexity contributed by each individual symbol or group of symb
ols. In particular, we apply this method to the mapping rules (symbol group
ings) commonly used in DNA sequence analysis. We find that strong-weak bind
ings are remarkable homogeneously distributed as compared to purine pyrimid
ine, and that A and T are the most heterogeneous distributed bases.