A method for making a model for the folded structure of a set of proteins from an evolutionary analysis of a set of aligned homologous protein sequences was claimed in Ser. No. 07/857,224. The instant application concerns methods for using these models. The first method is used to confirm or deny a hypothesis that two proteins are homologous, and is comprised of comparing a predicted structure model for one family of proteins with a predicted structure model for a second family of proteins, or an experimental structure for the second family, and deducing the presence or absence of homology based on the presence or absence of structural similarity flanking key residue motifs in the polypeptide sequence. The second method identifies mutations during the divergent evolution of a protein sequence that are potentially adaptive by identifying episodes during the divergent evolution of a family of proteins where there is a high absolute rate of amino acid substitution, or a high ratio of non-silent substitutions to non-silent substitutions. Amino acids that are changing during this episode are likely to be adaptive. The third is a method for identifying specific in vitro properties of the protein that are likely to play a physiological role in vivo in an organism. This methods involves synthesizing in the laboratory proteins having the reconstructed amino acid sequences of a protein before and after a period of rapid sequence evolution that characterizes adaptive substitution, measuring the in vitro properties of the protein before the episode of rapid sequence evolution, and then measuring the in vivo properties of the protein after the episode of rapid sequence evolution. The in vitro behaviors that remained unchanged through this episode are not likely to have adaptive significance physiologically. The in vitro behaviors that changed through this episode are likely to have adaptive significance physiologically. The fourth concerns method for organizing genome sized sequence databases.