This thesis develops a comparison scheme between statistical patterns and biological/biochemical patterns of biomacromolecules. We first label molecular sites according to certain categories of derived statistical patterns according to given biological and biochemical properties. The statistical scheme is general, can be interpreted meaningfully and applicable to small sample sequences. It categorizes the statistical patterns in a comprehensive way that generalizes invariant, conserved sites as well other types of statistical characteristics. The method is evaluated in experiments using a cancer-suppressor protein called the p53. The results show clear association patterns between characterized statistical patterns and biomolecular characteristics The findings of the analysis correspond closely to a recent study tracing the occurrences of cancers in human twins, thus confirming certain hereditary factors of cancers. An implication to analysis of biomolecules is that global biomolecular characteristics, as reflected in statistical patterns, play an important role in describing the functions of biomacromolecules at the molecular level and that local independent analysis, either using computational or laboratory techniques, is inadequate.