Comparative analysis of protein three-dimensional structures and an approach to the inverse folding problem

Comparative analyses of three-dimensional structures of proteins provide useful rules for protein structure prediction. Sequence templates that define the amino acid substitutions consistent with a known protein fold can be constructed using such rules. Such templates can define sequences with similar three-dimensional structures even where there are no significant sequence identities. In such cases protein structure may be predicted using an extension of the methods developed for comparative modelling of homologous proteins.