薛 定 谔Structure Based Virtual ScreeningApplications and Practical Guides段践辛段践辛 Senior Applications Scientist April 6th, 2010薛 定 谔薛 定 谔Drug Discovery to Drug Development PipelinePreexisting target Literature target Genomics ProteomicsIdentification HTS De novo design Virtual screening Optimization SAR studies In vivo efficacy ADMET studiesBioavailability studies ADME Long-term toxicity In vivo efficacy Synthesis scale-upEfficacy Bioavailability Dose Safety assessments Drug interactionsLead ID and OptimizationDevelopmentPh ITarget SelectionTime-consuming 2-5 yrs discovery; 13.2 yrs development Cost-intensive $1.1bn (1995-2000) - $1.7bn (2002-2002)1 Total R 33% (US)Ph IIPh III2. Data from CMR Compendium, (2002) PhRMA Survey (2002)1. Glibert . J. et al. (2003) In Vivo, the Business Drug Discovery Today; 2006; 11:580Before and after refinement with PrimeX薛 定 谔薛 定 谔Multiple Conformations of the TargetTarget may have multiple bioactive conformations Screening against one conformation will retrieve only actives to that conformation Other actives will be missed because of the rigid treatment of the receptor structure in docking Docking into multiple conformations of the target is an alternative Ensemble Docking薛 定 谔薛 定 谔Ensemble Screening against p38 MAP Kinase薛 定 谔薛 定 谔Treating Protein Flexibility (Induced Fit Docking)Ligand RMSD ( ) Target Receptor Ligand From: Rigid Receptor Docking Induced Fit Docking Aldose Reductase 2acr:_ 1ah3 6.5 0.9 Antibody DB3 1dba:H 1dbb 7.6 0.3 CDK2 1dm2:A 1aq1 6.2 0.8 CDK2 1aq1:_ 1dm2 0.6 0.8 CDK2 1buh:A 1dm2 6.4 1.1 COX-2 3pgh:A 1cx2 11.2 1.0 COX-2 1cx2:A 3pgh 6.6 1.0 (0.51) Estrogen Receptor 3ert:A 1err 2.3 1.4 (1.01) Estrogen Receptor 1err:A 3ert 5.3 1.0 Factor Xa 1ksn:A 1xka 9.3 1.5 Factor Xa1 1xka:C 1ksn 5.3 1.5 HIV-RT 1rth:A 1c1c 2.5 1.3 HIV-RT 1c1c:A 1rth 12.0 2.5 Neuraminidase 1nsc:A 1a4q 3.9 0.8 Neuraminidase 1a4q:A 1nsc 1.0 1.7 PPAR- 2prg:A 1fm9 9.8 3.0 (1.52) PPAR- 1fm9:D 2prg 9.1 1.8 (0.43) Thermolysin 1kr6:A 1kjo 1.1 1.3 Thermolysin 1kjo:A 1kr6 3.5 3.2 (1.64) Thymidine Kinase 1kim:A 1ki4 4.7 0.4 Thymidine Kinase 1ki4:A 1kim 0.5 1.2 Sherman, W.; Day, T.; Jacobson, M. P.; Friesner, R. A.; Farid, R., “Novel Procedure for Modeling Ligand/Receptor Induced Fit Effects“, J. Med. Chem. 49 (2006) 534 -553.薛 定 谔薛 定 谔Water in Binding SiteWater frequently mediate interactions between ligand and the protein. The decision to include or exclude water in the binding site is difficult and crucial. Compare available crystal structures and look for conserved water molecules. Validate by self-docking 薛 定 谔薛 定 谔Questions Before Virtual ScreeningWhich is my target? Literature search, biological validation Is my target druggable? SiteMap What are the characteristics of my binding pocket? SiteMap Is the target structure flexible? Induced Fit Docking What is the quality of the target structure? Electron density, PrimeX What are the side chain protonation states? Hydrogen bonding network? Protein preparation wizard Do I have multiple conformations of my target? Compare all PDB structures for the target, Maestro Water, water everywhere. Glide薛 定 谔薛 定 谔Questions Before Virtual Screening IIWhat is the source of my compound collection? Are the compounds in 3D? Do they have tautomers and ionization states generated? What is the typical size of compounds I am screening? Drug- like, lead-like, fragments? Is the compound collection diverse? Do the compounds have desired properties? How to filter the compounds?薛 定 谔薛 定 谔Ligand PreparationBinding sites may have different microscopic pH compared to the physiological pH of 7.4 Ionization states of ligands should be generated for larger pH range. Tautomerization will change the ligand hydrogen bonding capability. Quality geometry of bond length and angle is crucial because Glide samples only torsion angles. Proper preparation with ionization penalties will aid selection of hits.薛 定 谔薛 定 谔Running LigPrepAccept 2D structures or SMILES Use Epik for ionization If your target is metalloprotein, use Add metal binding states Stereoisomers薛 定 谔薛 定 谔Properties of Compounds - QikProp薛 定 谔薛 定 谔Filter the Compounds 薛 定 谔薛 定 谔Other Filtration AlternativesPharmacophore filtering Phase Based on known important interactions Shape based filtering Phase Shape Fast method without knowledge of important interactions 2D similarity filtering Canvas Very fast method without need to generate 3D structures Suitable for 10s of millions of compounds Fingerprint based QSAR based filtering Canvas Need activity data on many active molecules薛 定 谔薛 定 谔Diversity Analysis of Compound Collection薛 定 谔薛 定 谔Use Canvas to Select Diverse Compounds12薛 定 谔薛 定 谔Questions Before Virtual Screening IIWhat is the source of my compound collection? Compound vendor, ZINC database Are the compounds in 3D? Ligprep, Epik Do they have tautomers and ionization states generated? Epik What is the