際際滷

際際滷Share a Scribd company logo

Tautomers - Advanced Databases for in-silico Screening?

?
?
Frank Oellien
Frank Oellien

(1) Tautomers can be biologically relevant forms of ligands that interact with protein targets like bacteria, arthropods, and parasites. (2) Current virtual screening software does not fully account for tautomers in compound libraries and protein structures. (3) The author developed a tautomer enumeration approach using CACTVS to generate tautomers and conduct tautomer-sensitive duplicate checking for use in virtual screening workflows. (4) Examples on matrix metalloproteinase and cyclin-dependent kinase targets showed improved hit rates and separations when searching databases containing enumerated tautomers compared to databases without tautomers.

1 of 30
Download to read offline
Tautomers - Advanced Databases for in-silico Screening ? Frank Oellien, 18th CIC Workshop 2004, Boppard
Overview ? Motivation for tautomers in screening data sets ? Tautomer enumeration approach ? Workflow ? Examples for ｀enhanced¨ Database Search ? Results for ｀enhanced¨ Database Search ? Summary and Future Tasks
Biological Relevance of Tautomers Targets: Bacteria, Arthropods, and Parasites Variance for physiological pH expected C not static values !
Ambiguities in Proteins and Ligands Proteins (X-Ray structures) ? Flexibility (e.g. Gln, Asn) ? Ionisation states (e.g. Glu, Asp) ? Tautomerism (e.g. His) Ligands (Compound Libraries) ? ? ? ? Conformations Ionization states Stereo centers Tautomers
Software: Virtual Screening Types of Virtual Screening Software ? High-throughput Docking of ligands into protein X-Ray structures (Gold, FlexX) ? DB for pharmacophore search (Catalyst, Unity) Current VS software applications adress: ? ? ? ? Conformations Ionization states Stereo centers Tautomers ? X ? X (exception FlexX 2004)
Biological Relevance of Tautomers Tautomeric states of ligands can be relevant for biological interactions ? Derivates of tetrazole, triazole, thiazole, pyrazole, iminopyrimidine, ´ ? Brandstetter et al., MMP-8-Inhibitors J. Biol. Chem. 276, 2001, 17405-12. ? Pospisil et al., Ligands of herpesviral thymidine kinases, Helvet. Chim. Acta 85, 2002, 3237-50.
Software: Tautomer Generation Tautomer Generation Applications ? Agent 2.0 (ETH Zurich- Switzerland) ? QUAC PAC 1.1 (OpenEyes) ? StereoPlex (Tripos) ? no extensions by the means of user-defined rules ? no tautomer-sensitive duplicate check Aim: Easily extensible and scriptable software that allows the integration and automation of tautomer generation in our existing screening workflow. ? CACTVS: Chemical data management system
Tautomer enumeration ? C core library, Tcl command layer ? Main command: ens transform $eh $tlist <direction> <reactionmode> <flags> <overlapmode> <excludelist> <maxtautomers> <timeout> tlist: Transformation definition - SMIRKS line notation [#1:1][O:2][C:3]#[N:4]>>[O:2]=[C:3]=[N:4][#1:1] ? preferred tautomer forms ? tautomer sensitive duplicate check ? 21 pre-defined rules (up to 1,11-H-shifts) ? user-defined tautomer sets
Examples N N H2N N H1 N H N H2N N H2 H N HN N N H2N N N N N H 11 N HN N H2N N 4 H N N N H 12 HN HN N H N N H2N N 5 N HN 8 N H HN N N H 13 HN N 14 N H 10 N H OH N HN N N OH N HN N H OH N H9 N N N O OH H2N OH H N N H3 HN 7 O HN H2N N HN OH H N N 6 HN N N N OH O H2N H N OH O OH O O N H H N HN HN N N 15 Only 2 transformation rules are needed (1,3 and 1,5-H-shift)
Database Expansion 1400000 x 3,4 1200000 1000000 x 3,5 800000 x 3,2 600000 400000 x 3,6 x 3,6 x 3,0 x 2,7 200000 0 Maybridge no tautom ers Specs with tautom ers TimTec V itasM A sinex Platinum A sinex Gold ChemDiv
Tautomer Enumeration - Benchmark Platform: SGI Fuel R1400 / 600 MHz, 1 GB RAM Performance depends on ? nature of the compounds ? number of tautomers SupplierDB Compounds/min Multiplier Maybridge Screening > 150 2,5 Asinex Platinum > 250 2,9 VitasM (in-hose Stock) > 560 3 Tripos Leadscreen > 1400 2,2
Tautomeric Fingerprints a) Asinex (Platinum Collection) 100 40 30 60 20 10 40 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 20 b) Specs (Screening Collection) 100 0 1 4 7 10 13 16 20 24 28 34 No of tautomer forms 39 Tautomer frequency Tautomer frequency 50 80 46 80 52 69 50 115 40 30 60 20 10 40 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 20 0 1 9 17 25 33 41 49 57 65 74 84 95 108 125 149 182 210 348 742 No of tautomer forms
Virtual Screening Workflow @ Intervet 2D / 3D Structure DB (MDL) PreProcessing Tautomer Generation Specific 3D Databases (Catalyst, Unity) Tautomer-sensitive Duplicate check Data Analysis Virtual Screening
Example I - MMP-8 (PDB Entry Code: 1JJ9) Matrix Metalloproteinase Inhibitor: 8-Barbiturate H. Brandstetter et al., MMP-8Inhibitors J. Biol. Chem. 276, 2001, p 17405-12.
http://home.t-online.de/home/kubinyi/dd-18.pdf
Example I C MMP-8 Pharmacophore from X-RAY Matrix Metalloproteinase Inhibitor: 8-Barbiturate H-Bond Donor: green H-Bond Acceptor: magenta Hydrophobic aliphatic: blue Ring Aromatic: brown
Example I C MMP-8 Testdatabase 993 molecules selected from NCI 2000 Database (Catalyst Version) 2D / 3D Structure DB (NCI 2000) Divers Compound Selection (Cerius 2 4.9) 3D Database (Catalyst, No Tautomers) Tautomer Generation (CACTVS) 994 compounds including 8-Barbiturate (X-RAY) * 3D Database Default conditions for Catalyst database building C exception: max conformers 150 (Catalyst, with tautomers) 1536 compounds including 5 8-Barbiturate Tautomers * Knowledge of stable tautomeric form is needed as prerequisite.
Example I C MMP-8 Search on Testdatabase Questions: ? Covergage of the suggested tautomeric states for the 8barbiturate by our workflow ? ? Conformer generation of Catalyst resemble X-Ray ? ? Fit value significant for X-Ray and also conformers ? ? Signal to noise relationship C fit values of other hits ? 3D Database Exotic luxury or useful effort (Catalyst, WithTautomers) 1536 compounds including 5 8-Barbiturate Tautomers ? 3D Database (Catalyst, No Tautomers) 994 compounds including 8-Barbiturate (X-RAY)
Example I C MMP-8 Results Best Search C no tautomers: ? 29 compounds found ? 8-barbiture acid found in database, but scores less significant (BestFit 3.2) ? 8-barbiture tautomer form (X-ray) BestFit 7.2 (AV= 3.0 SD = 1.7) ? second best scored hit show BestFit 6.2 ? only 3 hits score higher then BestFit 4 ? best non X-Ray conformer scores BestFit 6.3
Example I C MMP-8 Results Best Search - with tautomers: ? 30 compounds found (22 unique, 8 tautomeric duplicates) ? 8-barbiture BestFit 7.2 (AV= 2.0 SD = 1.5) ? second best scored hit show BestFit 5.5 ? only 3 hits score higher then BestFit 4 ? non tautomeric 8-barbiture scores BestFit 3.2 ? 60 % Overlap between both search results (all top scoring hits in common)
Example I C MMP-8 Summary ? Significant better fit value and hit separation in case of a database search including tautomers ? X-ray structure closely resembled ? Number unique hits reduced ? Significant more structures have to be converted C time consuming aspect Critical aspect: ? Hit rate (unique hits) is lower for database including tautomers (?) ? X-ray structure or known physiological conditions in the protein appear to be important for sensitive pharmacophore searches
Example II - CDK of Eimeria tenella Cyclin Dependent Kinase C Homology Model based on Sequence Analysis C. Beyer et. al. Oral & Poster Presentation at the 18. Darmst?dter Molecular Modelling Workshop 2004 J.H. Kinnaird et al. International Journal for Parasitology 34, 2004, 683C692
Example II - CDK of Eimeria tenella - hiphop Qualitative pharmacophore model derived from human CDK2 best selective inhibitors C prefilter for docking libraries Feature mapping of pharmacophore hypothesis with CDK2 selective molecules H-Bond Donor: green H-Bond Acceptor: magenta Hydrophobic: blue Ring Aromatic: brown
Example II - CDK of Eimeria tenella - database 993 molecules selected from NCI 2000 Database (Catalyst Version) plus 123 known human CDK1/2 inhibitors. 93 ligands show activity against CDK2. 3D Database 3D Database (Catalyst, WithTautomers) (Catalyst, No Tautomers) 2368 compounds including 837 CDK1/2 inhibitor tautomers * *733 CDK2 inhibitor tautomers 1116 compounds including 123 CDK1/2 known inhibitors
Example II - CDK of Eimeria tenella - Results A) Search results without tautomers: ? best hypothesis finds 55.5 % of the CDK2 known ? Inhibitors (AV 39.8 % SD 9.5 %) ? Best selective (selectivity higher 4 included) ? Number of hits 81 B) Search results with tautomers: ? best hypothesis finds 72.2. % of the CDK2 known ? Inhibitors (AV 66.4 % SD 10.0 %) ? Best selective (selectivity higher 4 included) ? Number of unique hits 61 ? Overlap of best hypo search in A) with results in B) 93 %
Example II - CDK of Eimeria tenella - Results A) Search results without tautomers: ? GH Score 0.54 for best pharmacophore model B) Search results with tautomers: ? GH Score 0.77 for best pharmacophore model of A)
Example II - CDK of Eimeria tenella - Summary ? Number of true hits better in case of tautomers ? High overlap among top scoring ligands for both searches Remark: SBF models under way Critical aspect: ? Number of unique hits is reduced by using tautomer databases ? All kinds of tautomeric states are considered.
Future Tasks ? Modifications of tautomeric rules ? Automatisation of database building workflow ? Implementation of defined Ionisations ? Further investigations with examples
Acknowledgements Dr. J. Cramer, C. Bayer Dr. J. Schr?der, PD Dr. P. Selzer Intervet Innovation GmbH Dr. W.-D. Ihlenfeldt Χemistry GmbH Dr. O. Sacher Molecular Networks GmbH Dr. T. Hidaka Takeda Pharmaceutical Ltd.
Paul Selzer Richard Marh?fer Andreas Rohwer J?rg Schr?der J?rg Cramer BIOCHEMINFORMATICS Who we are ... Anette Klinger Carsten Beyer Frank Oellien Kristin Engels Andreas Krasky Hon Tran

More Related Content

Tautomers - Advanced Databases for in-silico Screening?

  • 1. Tautomers - Advanced Databases for in-silico Screening ? Frank Oellien, 18th CIC Workshop 2004, Boppard
  • 2. Overview ? Motivation for tautomers in screening data sets ? Tautomer enumeration approach ? Workflow ? Examples for ｀enhanced¨ Database Search ? Results for ｀enhanced¨ Database Search ? Summary and Future Tasks
  • 3. Biological Relevance of Tautomers Targets: Bacteria, Arthropods, and Parasites Variance for physiological pH expected C not static values !
  • 4. Ambiguities in Proteins and Ligands Proteins (X-Ray structures) ? Flexibility (e.g. Gln, Asn) ? Ionisation states (e.g. Glu, Asp) ? Tautomerism (e.g. His) Ligands (Compound Libraries) ? ? ? ? Conformations Ionization states Stereo centers Tautomers
  • 5. Software: Virtual Screening Types of Virtual Screening Software ? High-throughput Docking of ligands into protein X-Ray structures (Gold, FlexX) ? DB for pharmacophore search (Catalyst, Unity) Current VS software applications adress: ? ? ? ? Conformations Ionization states Stereo centers Tautomers ? X ? X (exception FlexX 2004)
  • 6. Biological Relevance of Tautomers Tautomeric states of ligands can be relevant for biological interactions ? Derivates of tetrazole, triazole, thiazole, pyrazole, iminopyrimidine, ´ ? Brandstetter et al., MMP-8-Inhibitors J. Biol. Chem. 276, 2001, 17405-12. ? Pospisil et al., Ligands of herpesviral thymidine kinases, Helvet. Chim. Acta 85, 2002, 3237-50.
  • 7. Software: Tautomer Generation Tautomer Generation Applications ? Agent 2.0 (ETH Zurich- Switzerland) ? QUAC PAC 1.1 (OpenEyes) ? StereoPlex (Tripos) ? no extensions by the means of user-defined rules ? no tautomer-sensitive duplicate check Aim: Easily extensible and scriptable software that allows the integration and automation of tautomer generation in our existing screening workflow. ? CACTVS: Chemical data management system
  • 8. Tautomer enumeration ? C core library, Tcl command layer ? Main command: ens transform $eh $tlist <direction> <reactionmode> <flags> <overlapmode> <excludelist> <maxtautomers> <timeout> tlist: Transformation definition - SMIRKS line notation [#1:1][O:2][C:3]#[N:4]>>[O:2]=[C:3]=[N:4][#1:1] ? preferred tautomer forms ? tautomer sensitive duplicate check ? 21 pre-defined rules (up to 1,11-H-shifts) ? user-defined tautomer sets
  • 9. Examples N N H2N N H1 N H N H2N N H2 H N HN N N H2N N N N N H 11 N HN N H2N N 4 H N N N H 12 HN HN N H N N H2N N 5 N HN 8 N H HN N N H 13 HN N 14 N H 10 N H OH N HN N N OH N HN N H OH N H9 N N N O OH H2N OH H N N H3 HN 7 O HN H2N N HN OH H N N 6 HN N N N OH O H2N H N OH O OH O O N H H N HN HN N N 15 Only 2 transformation rules are needed (1,3 and 1,5-H-shift)
  • 11. Tautomer Enumeration - Benchmark Platform: SGI Fuel R1400 / 600 MHz, 1 GB RAM Performance depends on ? nature of the compounds ? number of tautomers SupplierDB Compounds/min Multiplier Maybridge Screening > 150 2,5 Asinex Platinum > 250 2,9 VitasM (in-hose Stock) > 560 3 Tripos Leadscreen > 1400 2,2
  • 12. Tautomeric Fingerprints a) Asinex (Platinum Collection) 100 40 30 60 20 10 40 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 20 b) Specs (Screening Collection) 100 0 1 4 7 10 13 16 20 24 28 34 No of tautomer forms 39 Tautomer frequency Tautomer frequency 50 80 46 80 52 69 50 115 40 30 60 20 10 40 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 20 0 1 9 17 25 33 41 49 57 65 74 84 95 108 125 149 182 210 348 742 No of tautomer forms
  • 13. Virtual Screening Workflow @ Intervet 2D / 3D Structure DB (MDL) PreProcessing Tautomer Generation Specific 3D Databases (Catalyst, Unity) Tautomer-sensitive Duplicate check Data Analysis Virtual Screening
  • 14. Example I - MMP-8 (PDB Entry Code: 1JJ9) Matrix Metalloproteinase Inhibitor: 8-Barbiturate H. Brandstetter et al., MMP-8Inhibitors J. Biol. Chem. 276, 2001, p 17405-12.
  • 16. Example I C MMP-8 Pharmacophore from X-RAY Matrix Metalloproteinase Inhibitor: 8-Barbiturate H-Bond Donor: green H-Bond Acceptor: magenta Hydrophobic aliphatic: blue Ring Aromatic: brown
  • 17. Example I C MMP-8 Testdatabase 993 molecules selected from NCI 2000 Database (Catalyst Version) 2D / 3D Structure DB (NCI 2000) Divers Compound Selection (Cerius 2 4.9) 3D Database (Catalyst, No Tautomers) Tautomer Generation (CACTVS) 994 compounds including 8-Barbiturate (X-RAY) * 3D Database Default conditions for Catalyst database building C exception: max conformers 150 (Catalyst, with tautomers) 1536 compounds including 5 8-Barbiturate Tautomers * Knowledge of stable tautomeric form is needed as prerequisite.
  • 18. Example I C MMP-8 Search on Testdatabase Questions: ? Covergage of the suggested tautomeric states for the 8barbiturate by our workflow ? ? Conformer generation of Catalyst resemble X-Ray ? ? Fit value significant for X-Ray and also conformers ? ? Signal to noise relationship C fit values of other hits ? 3D Database Exotic luxury or useful effort (Catalyst, WithTautomers) 1536 compounds including 5 8-Barbiturate Tautomers ? 3D Database (Catalyst, No Tautomers) 994 compounds including 8-Barbiturate (X-RAY)
  • 19. Example I C MMP-8 Results Best Search C no tautomers: ? 29 compounds found ? 8-barbiture acid found in database, but scores less significant (BestFit 3.2) ? 8-barbiture tautomer form (X-ray) BestFit 7.2 (AV= 3.0 SD = 1.7) ? second best scored hit show BestFit 6.2 ? only 3 hits score higher then BestFit 4 ? best non X-Ray conformer scores BestFit 6.3
  • 20. Example I C MMP-8 Results Best Search - with tautomers: ? 30 compounds found (22 unique, 8 tautomeric duplicates) ? 8-barbiture BestFit 7.2 (AV= 2.0 SD = 1.5) ? second best scored hit show BestFit 5.5 ? only 3 hits score higher then BestFit 4 ? non tautomeric 8-barbiture scores BestFit 3.2 ? 60 % Overlap between both search results (all top scoring hits in common)
  • 21. Example I C MMP-8 Summary ? Significant better fit value and hit separation in case of a database search including tautomers ? X-ray structure closely resembled ? Number unique hits reduced ? Significant more structures have to be converted C time consuming aspect Critical aspect: ? Hit rate (unique hits) is lower for database including tautomers (?) ? X-ray structure or known physiological conditions in the protein appear to be important for sensitive pharmacophore searches
  • 22. Example II - CDK of Eimeria tenella Cyclin Dependent Kinase C Homology Model based on Sequence Analysis C. Beyer et. al. Oral & Poster Presentation at the 18. Darmst?dter Molecular Modelling Workshop 2004 J.H. Kinnaird et al. International Journal for Parasitology 34, 2004, 683C692
  • 23. Example II - CDK of Eimeria tenella - hiphop Qualitative pharmacophore model derived from human CDK2 best selective inhibitors C prefilter for docking libraries Feature mapping of pharmacophore hypothesis with CDK2 selective molecules H-Bond Donor: green H-Bond Acceptor: magenta Hydrophobic: blue Ring Aromatic: brown
  • 24. Example II - CDK of Eimeria tenella - database 993 molecules selected from NCI 2000 Database (Catalyst Version) plus 123 known human CDK1/2 inhibitors. 93 ligands show activity against CDK2. 3D Database 3D Database (Catalyst, WithTautomers) (Catalyst, No Tautomers) 2368 compounds including 837 CDK1/2 inhibitor tautomers * *733 CDK2 inhibitor tautomers 1116 compounds including 123 CDK1/2 known inhibitors
  • 25. Example II - CDK of Eimeria tenella - Results A) Search results without tautomers: ? best hypothesis finds 55.5 % of the CDK2 known ? Inhibitors (AV 39.8 % SD 9.5 %) ? Best selective (selectivity higher 4 included) ? Number of hits 81 B) Search results with tautomers: ? best hypothesis finds 72.2. % of the CDK2 known ? Inhibitors (AV 66.4 % SD 10.0 %) ? Best selective (selectivity higher 4 included) ? Number of unique hits 61 ? Overlap of best hypo search in A) with results in B) 93 %
  • 26. Example II - CDK of Eimeria tenella - Results A) Search results without tautomers: ? GH Score 0.54 for best pharmacophore model B) Search results with tautomers: ? GH Score 0.77 for best pharmacophore model of A)
  • 27. Example II - CDK of Eimeria tenella - Summary ? Number of true hits better in case of tautomers ? High overlap among top scoring ligands for both searches Remark: SBF models under way Critical aspect: ? Number of unique hits is reduced by using tautomer databases ? All kinds of tautomeric states are considered.
  • 28. Future Tasks ? Modifications of tautomeric rules ? Automatisation of database building workflow ? Implementation of defined Ionisations ? Further investigations with examples
  • 29. Acknowledgements Dr. J. Cramer, C. Bayer Dr. J. Schr?der, PD Dr. P. Selzer Intervet Innovation GmbH Dr. W.-D. Ihlenfeldt Χemistry GmbH Dr. O. Sacher Molecular Networks GmbH Dr. T. Hidaka Takeda Pharmaceutical Ltd.
  • 30. Paul Selzer Richard Marh?fer Andreas Rohwer J?rg Schr?der J?rg Cramer BIOCHEMINFORMATICS Who we are ... Anette Klinger Carsten Beyer Frank Oellien Kristin Engels Andreas Krasky Hon Tran