Drug Target (Study on computational method of discovering new target in drug discovery) :. 2006 6 26 2006 8
,,,. 12 9.. target target..,. in silico. drug, disease, target (PharmDB) (phexplorer). drug, disease, target,. drug, disease, target, target drug target. target. target target 1% 43%. :,,,, Java : 99448-099
1. 1. 2. 2. 1. Network 2. Pajek & Prefuse 3. Cell Signaling and Motif 3. phexplorer - Visualization Tool 4. Discovery of new target 5. 6. 7.
List of Figures Figure 1. Pajek Figure 2. Pajek Figure 3. Prefuse Figure 4. Signaling network of mammalian hippocampal CA1 neuron Figure 5. Network Motif Figure 6. Map of regulatory profile within cellular network Figure 7. http://pharmdb.org Figure 8. Drug-Target-Disease Network Figure 9. phexplorer Figure 10. Popup Menu on Target Figure 11. Expansion Figure 12. DST Method Figure 13. DDST Method Figure 14. Related Targets of Acetaminophen Figure 15. Related Diseases of Acetaminophen Figure 16. Targets of Related Diseases Figure 17. Shared Targets of Releated Diseases Figure 18. Result of Brute-Force Method Figure 19. Result of DST Method - Test 1 Figure 20. Result of DST Method - Test 2 Figure 21. Result of DST Method - Test 3 Figure 22. Result of DDST Method - Test 1 Figure 23. Result of DDST Method - Test 2 Figure 24. Result of DDST Method - Test 3 Figure 25. Result of CS Method - Test 1 Figure 26. Result of CS Method - Test 2 Figure 27. Result of CS Method - Test 3 Figure 28. Propose a new Target Figure 29. Input Dialog Figure 30. Proposed Target
1. 1. (drug discovery) target Target Identification, target Hit Identification, Lead Identification, receptor Lead Optimization, Candidate Drug Prenomination, 1 Concept Testing, 2, 3 Development For Launch, Launch Phase, 4 ( ) Life Cycle Support. 12, 9. 100 1... drug target. drug target. irgb PharmDB (http://pharmdb.org) target. 2...,.. Mycoplasma genitalium E-Cell, Health center Virtual Cell
, KEGG,, NCBI Entrez. target drug, disease Therapeutic Target Databse(TTD) Protein target Nucleic Acid target synonym, disease drug, function target. Drug Bank drug drug target, Java structural searching. brute force drug-target matching silico. PharmDB phexplorer target drug target PharmDB in silico.
2. 1. Network Network( ) graph data N = (V,L,P,W). A graph g = ( V, L) V vertex, A arc, E edge L = E A. P: Vertex value function W: Line value function (directed, undirected, mixed). 2-mode networks Multi-relational networks Temporal networks Petri s nets, p-graphs, 2. Pajek & Prefuse Pajek (Figure 1) visualize. social network large scale. Prefuse (Figure 3) visualization tool Pajek. 3. Cell Signaling and Motif Motif. A. Ma ayan[5] 545 ligand & protein node 1259 function & biochemical mechanism link Mammalian hippocampal CA1 neuron signaling network Motif Motif cell signaling. MotifMFinder Program[6]. Motif local density interconnectedness DIP(Density of Information Processing) DIP.
Figure 1. Pajek Figure 2. Pajek
Figure 3. Prefuse
, Motif concentration MLI(Motif Location Index). MLI. cellular network regulatory profile map Figure 6. positive feedback loop negative feedback loop.
Figure 4. Signaling network of mammalian hippocampal CA1 neuron Figure 5. Network Motif
Figure 6. Map of regulatory profile within cellular network
3. phexplorer - Visualization Tool PharmDB drug, target, disease phexplorer. TTD DrugBank drug target, disease. Motif pathway. phexplorer Java network visualization library prefuse[2] PharmDB drug target, disease. drug-target-disease target target protein-protein interaction, target BLAST search homology research. PharmDB (Figure 7) Target Name Drug Name, Disease Name keyword GraphView Figure 8 drug-disease-target. Mouseover. Explorer Explorer. Antialiasing phexplorer speed Java applet (Linux MacOS Unix ). Explorer mode overview. target, drug, disease target drug, disease. Target Figure 10 Search BLAST search target protein-protein interaction target. expansion phexplorer Figure 11. Figure 11 ( ),, drug-targetdisease mark/unmark. phexplorer phexplorer. ( ) BLAST search. Self protein-protein interaction. protein-protein interaction edge
Figure 7. http://pharmdb.org Figure 8. Drug-Target-Disease Network
Figure 9. phexplorer Figure 10. Popup Menu on Target
Figure 11. Expansion
4. Discovery of new target phexplorer drug target. drug disease. disease target disease drug. target. phexplorer Acetaminophentarget Figure 14. Acetaminophen drug Figure 15. Figure 15 Acetaminophen Cyclogenase-3 target drug Pain Unspecified Inflammation. disease target Figure 16. disease Mark Delete target Figure 17 Cyclooxygenase-3, Peroxisome proliferator activated receptor gamma, Inhibitor of nuclear factor kappa-b kinase. Acetaminophen target, Cyclooxygenase-3 Acetaminophen drug target.. 1. UMAD 1 disease target (DST) DST(Disease Shared Target) method target drug disease target target. Different diseases and same target UMAD Disease Disease Disease Disease Target Target Target Target Target Figure 12. DST Method
2. UMAD disease drug target (DDST) DDST(Disease Drug Shared Target) target drug disease drug target target. DST, DDST. Figure 13. DDST Method targettarget disease drug. target drug UMAD target target,. PharmDB MySQL, mysqlpp 2.0.6 DB PharmDB drug list. drug target, disease. PharmDB. 2 PharmDB target : 2830
PharmDB disease : 811 PharmDB drug : 1192 target drug: 260 disease drug: 641 551 drug 811 disease, 2830 target
Figure 14. Related Targets of Acetaminophen Figure 15. Related Diseases of Acetaminophen
Figure 16. Targets of Related Diseases Figure 17. Shared Targets of Releated Diseases
5. target target disease drug(degree). target. degree. Test 1, Test 2, Test 3. Test 1: Degree 1 Test 2: Degree 2 Test 3: Degree 3 Brute-force Method target drug target target. PharmDB target target (p1). drug p1. n( Target) p 1 = n( Target) 1. 2830 target target Figure 18. Result of Brute-Force Method
Brute-force method target 0. DST DDST target target (p2). drug p2. n( Target Target) p2 = n( Target) DST (Disease-Shared-Target) method 1. Test 1 target Figure 19. Result of DST Method - Test 1 2. Test 2 target
Figure 20. Result of DST Method - Test 2 3. Test 3 target Figure 21. Result of DST Method - Test 3 DST method Test 1 Test 3 target.
DDST (Disease-Drug-Shared-Target) method 1. Test 1 target Figure 22. Result of DDST Method - Test 1 2. Test 2 target Figure 23. Result of DDST Method - Test 2 3. Test 3 target
Figure 24. Result of DDST Method - Test 3 DDST method Test1 Test3. Combined search DST DDST. p2 DST DDST. 1. Test 1 target
Figure 25. Result of CS Method - Test 1 2. Test 2 target Figure 26. Result of CS Method - Test 2 3. Test 3 target
Figure 27. Result of CS Method - Test 3 Combined search method target. DDST method DDST method target target.
6. DST method target. DDST method DST method target drug target target. Combined search DDST method. target degree 1, 2, 3 target. target target. phexplorer drug Propose New Target. Drug Figure 28. method Test 1, Test 2, Test 3 target degree (Figure 29). target degree Filter Value. phexplorer target. (Figure 30) PharmDB drug, target, disease drug target. target 80~90% target PharmDB target. DDST method combined search method disease degree, disease degree target. BLAST protein-protein interaction target.
Figure 28. Propose a new Target Figure 29. Input Dialog
Figure 30. Proposed Target
7. [1] Batagelj, V. and A. Mrvar, Pajek: Analysis and Visualization of Large Networks, in Graph Drawing Software, Springer, p. 77-103, 2003. [2] Jeffrey Heer, prefuse: a software framework for interactive information visualization, in Masters of Science, Computer Science Division, University of California, Berkeley, 2004 [3] Gyuman Yoon, Kyoohyung Rho, Philip M. Kim, Byungnam Kahng, Sunghoon Kim, Systematic analysis of commercial drugs, biological targets, and related diseases using network theory, in Satellite Meeting of STATPHY 22, 2004 [4], Biocomplex networks: identification of essential genes and characterization of protein regulation networks = :,, 2006 [5] A. Ma ayan, S. L. Jenkins, S. Neves, A. Hasseldine, E. Grace, B. Dubin-Thaler, N. J. Eungdamrong, G. Weng, P. T. Ram, J. J. Rice, A. Kershenbaum, G. A. Stolovitzky, R. D. Blitzer, R. Iyengar, Formation of regulatory patterns during signal propagation in a mammalian cellular network, Science 309, 1078-1083, 2005 [6] M. Barrios-Rodiles, K. R. Brown, B. Ozdamar, R. Bose, Z. Liu, R. S. Donovan, F. Shinjo, Y. Liu, J. Dembowy, I. W. Taylor, V. Luga, N. Przulj, M. Robinson, H. Suzuki, Y. Hayashizaki, I. Jurisica, J. L. Wrana, High-throughput mapping of a dynamic signaling network in mammalian cells, Science 307, 1621-1625, 2005 [7], = (A)Development of modeling system through a development of model and data integration methodology of systems biology, KAIST,, 2005 [8] K. Lindpaintner, Genetics and genemoics: impact on drug discovery and development, in Proceedings of the Fifth Annual international Conference on Computational Biology (Montreal, Quebec, Canada, April 22-25, 2001). RECOMB '01. ACM Press, New York, NY, 221-222, 2001. [9] M. Kashtan, S. Itzkovitz, R. Milo, U. Alon, Bioinformatics 20, 1746, 2004
Abstract Developing a new drug is one of the most time consuming and expensive field of science. Moreover, it requires experts from multi-disciplinary fields of biology, chemistry, biochemistry and molecular biology. On average, it takes 12 years and costs 9 billion dollars in order to develop one drug, and even after the development, there is a risk of not getting approved for sale by the Food and Drug Administration (FDA) in the United States. For pharmaceutical companies, it is a huge risk to invest that much time and money to develop one drug. However, with advancement of computer science and engineering, many research groups are attempting to develop in silico algorithms or tools (software) to narrow down candidate drugs or better understand the mechanism of disease. Doing so would reduce the development time and save money significantly. In this thesis, a computational methodology for finding new targets of existing drugs is discussed. First, network analysis tool (phexplorer) is developed based on the drug, disease and target database (PharmDB). phexplorer represents drug, disease and target relationships as an undirected graph; each node can be expanded and deleted by users. Then, a new algorithm is proposed to find new targets of drug from drug, disease and target relationships using phexplorer. Results show that the algorithm finds new targets better than the brute-force method. The brute-force method shows 1% of discovery ratio but the proposed algorithm shows up to 43% of discovery ratio. Keyword: Bioinformatics, Drug discovery, Bio information database, Interacting Network, Java Student number: 99448-099