Parametric Templates: A New Enzyme Active-Site Prediction Algorithm

It is an important problem to find functionally analogous enzymes based on the local structures of active-sites. Conventional methods predict active-sites by computing the deviations from the local-structure templates with no statistical parameters. We present a new statistical algorithm that uses parametric templates to compute the deviations of local sites. The parameters of the templates are determined automatically from a set of known active-sites. In this work, promising experimental results are shown through comparison of parametric templates with conventional templates.