UCLA-DOE Laboratory of Structural Biology and Molecular Medicine, Box 951570, UCLA, Los Angeles CA 90095
To gain a full understanding of the molecular biology of organisms, we must be able to assign genome sequences to 3D structures and functions. Fold assignment based on comparison of amino acid sequences works well when the new sequence can be aligned to a sequence of a known 3D fold with an identity of 25% or more between corresponding residues. For more distant relationships, profile-threading methods are helpful in making additional fold assignments.
Our current work is aimed at advancing structure-based fold assignment methods in several ways. One is to use the predicted secondary structure of the sequence as an aid in assigning the sequence to one of a library of known folds. Another method is the directional profile. The directional profile takes account of the distribution in space of atoms around each residue in a structure, and defines the library of known folds in terms of these distributions.
In tests on the genome of mycoplasm genitalium, structure-based fold assignment methods are able to assign 15-20% more sequences to folds than are sequence-sequence methods, with most of the additional assignments representing very distant structure-sequence relationships. For such work, we have set up a network-accessible server for fold assignments: the URL is: http://www.mbi.ucla.edu/people/frsvr/frsvr.html. A second network-accessible server provides verification of protein models, based on the capatibility of each residue for its local environment in the protein. The URL of this server is: http://www.mbi.ucla.edu/verify3d.html