| Methods for determining the degree of similarity between groups of sequences have become important tools for understanding the relationships between proteins. Multiple sequence alignments in particular reveal the important amino acid positions within protein families. Sequence motifs such as position-specific scoring matrices, PROSITE patterns, and more recently, e MOTIFs add to the value of multiple sequence alignments by distilling out the conserved positions and providing simple expressions for searching databases for new family members.; Sequence motifs alone however, only reveal the positions that are evolutionarily conserved within a family; they do not reveal the functional or structural reasons for their conservation. The work presented in this dissertation helps bridge this gap, providing structural and functional clues for the conserved amino acids discovered in sequence motifs. Part I describes the development of 3MOTIF (http://3motifstanford.edu/) and 3MATRIX (http://3matrix.stanford.edu/), two software packages for visualizing conserved protein sequence motifs. These tools map several databases of discrete and probabilistic motifs onto the protein structure database, allowing researchers to visualize motifs in their appropriate three-dimensional context. Additional properties such as conservation strength and solvent accessible surface area are also mapped onto the conserved motifs, further indicating the possible functions of the conserved amino acids.; Part II describes two large-scale structural studies of the eMOTIF database of discrete sequence motifs. In these studies, it was shown that most eMOTIFs are structurally conserved as measured by structural alignment. Further, structures in which the same eMOTIF appear were shown to be members of the same SCOP domain in the majority of cases, indicating that eMOTIFs are useful predictors of the overall structural classification of proteins. In addition, the small fraction of eMOTIFs found in proteins of different SCOP classification were often observed to bind small molecules or cofactors such as ATP and NAD/FAD, molecules that are important in many different types of proteins. Preliminary work also shows that a significant number of eMOTIFs interact structurally, indicating that conserved sequence elements that are distant in sequence can actually represent the same evolutionarily conserved unit in the final folded structure. |
