Project 19: Classification of protein structure motif conformations Organization: Queen's University, Computing and Abstract Information Science Primary Contact : Janice Glasgow Similar to the concept of scene analysis in vision, Address : Department of Computing and Information molecular scene analysis is concerned with the Science processes of reconstruction and classification of complex molecular images. Such analyses rely on the Queen's University Kingston, Ontario availability of a priori information in the form of Canada, K7L 3N6 protein structure motifs. Researclh in this project has Email: janice@qucis.queensu.ca involved the study, development and implementation URL: http://www.qucis.queensu, ca/MSA/index.html of data mining tools for obtaining such structural motifs. The rapid growth of protein databases has created a demand for novel and efficient techniques for Duration: 10 years the classification of motif conformations which can be Number of People: 12 used to anticipate or predict potential substructures in Tools Developed: IMEM, Coincidence Detection protein images based on primary sequence Academic Disciplines: computing, biochemistry, information. chemistry, and mathematics Funding Sources: NSERC, CITO, IRIS, and PENCE Keywords: protein structure motif, molecular scene analysis, bioinformatics Project Related Publications: Glasgow, J.I., Steeg, E. and Fortier, S., "Motif Discovery in Protein Structure Database", In Pattern Discovery in Molecular Biology: Tools, Techniques andApplications, Wang, Shapiro and Shasha (Eds.), Oxford Press, (To appear). Conklin, D., Forlier, S., Glasgow, J.I. and Allen, F.H., "Conformational Analysis from Crystallographic Data using Conceptual Clustering", Acta Crystallographica, B52, 535-549,1996. Project 20: Similarity Measures o f R N A structures Organization: German National Research Center for Abstract Information Technology Primary Contact: Tamas Horvath Untranslated regions of (pre-) mRNAs contain Address: GMD - SET.KI regulatory elements called signal structures to control Sclfloss Birlinghoven gene expression at the posttranscriptional level. Their 53754 Sankt Augnstin regulatory information is encoded in their primary and Germany secondary structures. Given a set of known signal Email: tamas.horvath@gmd.de structures belonging to several classes, our aim is to URL: find a) the presence of signal structures belonging to http://www-fit-ki, gmd.de/projects/ml/relational-ibl.html known classes on uncharacterized mRNA, and b) uncharacterized signal structure classes. We have applied relational instance-based method for the first Duration: 5 years problem. Number of People: 6 Tools Developed: RIBL Keywords: mRNA signal structures, machine learning Academic Disciplines: machine learning, inductive logic programming Project Related Publications: U.Bohnebeck, T.Horvath, and S.Wrobel, "Term comparisons in First-order Similarity Measures", In D. Page, editor, Proceedings of the 8th International ConJkrence on Inductive Logic Programming, LNAI 1446, 65-79. SpringerVerlag, 1998. U.Bohnebeck, W.Salter, T.Horvath, S.Wrobel, and D.Blohm, "Measuring Similarity of RNA Structures by Relational Instance-Based Learning: A First Step Toward Detecting RNA Signal Structures in Silico", (accepted to GCB'98

/lp/association-for-computing-machinery/similarity-measures-of-rna-structures-0UJyt05K09