Sequence similarity scores and the inference of structure-function relationships

Improved methods are described for the interpretation of two or more aligned protein or nucleic acid sequences. These methods can be used to interpret the possible biological importance of regions within a known three-dimensional structure, or, even without a structure, to correlate sequence similarity with the known function of particular amino acids and to associate sequence similarity with properties predicted from the sequences. Improvements include the calculation of a position-dependent, gap-penalized similarity score; computer-assisted graphical association of sequence similarity with structural, functional or chemical properties of the sequences; and statistical comparisons of the sequence conservation or variability of different groups of residues. An application is described to analyze the sequences of piconarviral capsid proteins.