46 - Analysing Protein Families: Sequence, Structure, and Specificity of the SH2 Domain
Bissan Al-Lazikani, Felix Sheinerman, Barry Honig, Columbia University
We combined various sequence and structure analysis methods to analyse the diverse SH2 family. As well as gaining substantial insights into the biology of SH2 domains, an effective suite of sequence and structure analysis tools was developed that can be applied to study any diverse protein family.

47 - Sequence Signals for Generation of Antigenic Peptides by the Proteasome
Yael Altuvia, Hanah Margalit, The Hebrew University
Analysis of the termini and flanking regions of peptides eluted from MHC class I molecules suggests that the C-terminus and its immediate flanking position possess proteasomal cleavage signals. These signals can be used to assess the cleavage potential of peptides and aid in discrimination between immunodominant and cryptic peptides.

48 - Multiple Sequence Alignment and Homology Modeling of Sulfotransferase Enzymes
Keith W. Burdick, Irwin D. Kuntz, University of California, San Francisco
Members of the sulfotransferase superfamily of enzymes catalyze the transfer of a sulfuryl group from 3'-phosphoadenosine 5'-phosphosulfate to a wide variety of substrates. The global sequence similarity between the families within sulfotransferases is low (14-20%), making it difficult to align sequences. The crystal structure of estrogen sulfotransferase was used as a template in modeling the nucleotide binding site of four sulfotransferases.

49 - Identifying the DNA Binding Specificity of Hoxb-3
C. N. L. Chan, L. M. Jakt, M. H. Sham, D. K. Smith, University of Hong Kong
Hox proteins are mammalian transcription factors involved in antero-posterior axial identity in embryonic development. To identify the DNA binding specificity of Hoxb-3, chick and mouse genomic DNA were fragmented and the fragments were bound by murine Hoxb-3. Chi-square test was applied to find out the consensus binding sites and TCATTAATTGGC is proposed.

50 - Protein Name Tagging
Berry de Bruijn, Joel Martin, National Research Council, Canada
Automatic tagging of protein names is one component of our system for information extraction from biomedical articles. Our tagger combines three approaches: 1) analysis of string morphology; 2) context analysis; and 3) dictionary lookup. An evaluation compares our tagger with expert performance and a competing system.

51 - Classification of the Short-chain Dehydrogenase/Reductase Family
Yvonne Kallberg, Johan Nilsson, Bengt Persson, Karolinska Institutet, Sweden
We have classified the short-chain dehydrogenase/reductase (SDR) family of proteins by applying and comparing methods for protein homology search and subfamily classification. The low residue identity between some of the protein sequences (15-30%) and the large number of family members (~1500) provides a true challenge for the techniques investigated.

52 - Geometric Interpretation of Homologous Protein Families
Boris Kiryutin, Roman L. Tatusov, National Center for Biotechnology Information
Interpreting the similarity scores as scalar products allows the representation of biological sequences as vectors in multidimensional space. The eigenvectors technique produces the coordinates of these vectors in the most characteristic low dimensional subspace. The eigenvector approach was applied to Clusters of Orthologous Groups (COGS) as well as to complete genomes.

53 - Mechanism and Benefits of Catalytic Inefficiency in G-proteins
Mickey Kosloff, The Hebrew University
The G-protein family act as molecular switches in many signalling cascades. They are turned "on" by exchange of bound GDP to GTP and turned "off" by the intrinsic GTPase reaction. We analyze this inefficient catalysis using available crystal structures combined with functional analysis. We propose a model for the rate-limiting step of GTPase and discuss its implications for catalysis and biological function.

54 - Protein Set Analysis with SRS
David P. Kreil, T. M. Etzold, Wellcome Trust Genome Campus, UK
We present selected features of a survey for homopolymeric peptides (amino acid ‘runs’) in SRS-extracted protein sets. For example, a puzzling lack of asparagine runs longer than five residues in vertebrate proteins was discovered. This is in marked contrast to the many long glutamine runs, considering the high similarity of the residues.

55 - Maximum Likelihood Methods Reveal Conservation of Function among Closely Related Kinesin Families
Carolyn J. Lawrence, Russell L. Malmberg, University of Georgia; Michael G. Muszynski, Pioneer Hi-Bred International; R. Kelly Dawe, University of Georgia
We used progressive and HMM alignment methods along with parsimony, NJ, and ML treebuilding methods to construct a phylogeny of the kinesin superfamily. Intron positions are compared. Organismal and kinesin phylogenies are reconciled. A method that determines which regions of an alignment contribute to long branch attraction is described.

56 - GTC birdsEye: An Integrated Visualization Tool for Protein-protein Interaction Analysis
Anzhi Li, Veronique Damagnez, Genome Therapeutics Corporation
An integrated visualization tool, GTC birdsEye, has been developed for Y2H functional genomics to view the alignment patterns and binding domains of prey. GTC birdsEye supports the Y2H system to identify and characterize protein-protein interactions, and further extend the protein-protein interaction network.

57 - Evolutionary Trace Analysis as a Tool for Functional Analysis of Protein Structures.
Mike Marsh, Olivier Lichtarge, Baylor College of Medicine
Recent support for a structural genomics initiative will provide a more complete understanding of the universe of protein structures. Like large-scale sequencing projects pioneered last decade, however, the earliest stage result will be an accumulation of unannotated data. We show the use of the Evolutionary Trace (ET) method as a useful and general means of annotating protein structures for which there is little prior functional knowledge or mutational data.

58 - Studies on Short-chain and Medium-chain Dehydrogenases/Reductases Using Hitherto Completed Genomes
Bengt Persson, Jan-Olov Höög, Hans Jörnvall, Karolinska Institutet, Sweden
The protein families SDR and MDR constitute large enzyme families. Studies on these in the completed genomes reveal different evolutionary patterns for different sub-families, reflecting their different roles in the metabolism. A sub-class of SDR with dehydratase, epimerase, and isomerase activities seems to be of critical importance.

59 - Classifying Protein 3-D Structures by Point Set Pattern Matching Algorithms
Grace S. Shieh, Ming-Jing Hwang, Academia Sinica, Taiwan
A statistical measure is proposed to classify the 3-D structures of proteins. The measure captures the "similarity" between two protein sequences via an algorithm adapted from "Point set pattern matching (PSPM) in 3-dimension" (de Rezende and Lee, 1995) and "Faster PSPM in 3D" (Boxer, 1998).

60 - Computational Study on the Basis of Drug Resistance of the HIV-1 Protease
Wei Wang, Peter A. Kollman, University of California, San Francisco
We have studied several drug-resistant mutants of the HIV-1 protease using molecular dynamics and continuum solvent model. Several crucial residues have been identified for causing the resistance if they are mutated. The computational study has suggested several strategies for designing new inhibitors to combat the drug resistance.

61 - Large-scale Annotation of Functional and Structural Protein Sites
Allison Waugh, Glenn Williams, Russ B. Altman, Stanford University
The FEATURE system automates annotation of a query 3-D protein structure, predicting locations of functional and structural sites of interest. We are employing Legion, the University of Virginia's 'Worldwide Virtual Computer,’ to scan the Protein Data Bank for sites in FEATURE's site library, such as calcium binding, chloride binding, and redoxin active sites.