User Manual for the Danish (Q)SAR Database
Contents
1. e Search on all contained QSAR predictions and training set data e Combination of search results to make complex AND OR and NOT algorithms e Download of QSAR predictions in an RTF format document compatible with Microsoft Word and OpenOffice e Sorting on chemical similarity to facilitate read across groupings Launching the Danish QSAR Database Type in the following link in the address bar of the web browser http qsar food dtu dk DTU Food DTU norden Ministry of Environment and Food National Food The Danish Environmental Protection Agency 2 oe Nordic Council of Ministers Institute Danish Q SAR Database The database is developed by the N j i and the European Chemicals Agency micas Bp Leadscope sciMatics CS asis ACD Labs Figure 1 Opening screen for the Danish QSAR Database To begin searching click the Search button and the screen shown in Figure 2 should appear Click the button marked I agree to enter the database New search Searches Results Substances Structure PhysChem A Copyright notice terms and conditions of use ADME Permission is granted to use information from the database as is The database is an expert tool where the final assessment of properties is not dictated by the Q SAR estimates but by the user s own scientific judgment Aside from the fact Environment that models are never perfect the Q SAR field is under rapid development and models are re2. 72 000 are included in the structure set In addition chemical structures from other relevant databases are included leading to the new structure set of more than 600 000 unique chemical structures When possible the endpoints have been modelled in the three software systems Leadscope CASE Ultra and SciQSAR All DTU in house models and a number of commercial models have with the kind permission from MultiCASE been modelled in two or three systems The structure set has been predicted in the different systems and an overall battery prediction is made With the battery approach it is in many cases possible to reduce noise from the individual model estimates and thereby improve accuracy and or broaden the applicability domain All applied DTU QSAR models are documented in QMRFs QSAR Model Reporting Format Permissions to publish predictions for more than 600 000 substances were kindly provided by MultiCASE Inc Leadscope Inc Scimatics ACD Labs and US EPA The published predictions are abbreviated predictions simple yes no and do not include detailed information about specific alerts identified Applicability domain calls are however available Main features at a glance e Estimates for more than 600 000 chemicals in over 200 QSAR models e Contains experimental training set data for DTU models for which data are public e Search on substance ID and affiliation e Structure search on 2D structures as substructure or exact match
3. Page 1 Previous Next 1 Structures 1 1 of 1 Structure Id Similarity 21 Appendix 2 Searching by structure and similarity Example 1 Substructure search Search for molecules containing a fluorobenzene fragment It is possible to search for molecules which contain specific molecular fragments Start by clicking the Structure button to the left on the main search screen This will open the edit box shown below aer aalaleeja same Espia E q do a 0 2 Substructure search 24 Em G S ht 3 2 Exact match search al 4 7 boa 4 A Cancel In this example we will use the drawing tools in the edit box to create the fluorobenzene fragment Alternatively the fragment can be imported as a SMILES MOL file or from the chemical dictionary When drawing a fragment the different atoms bonds etc are added one by one Start by adding the benzene part of the molecule left click on the benzene button and then click on the blank canvas The canvas will look like this A O Rs B Ele SJ OJOJBI VJ OJ 9 19114 OS SJ JJ e llelo 27 Now click on the single bond button and then click on one of the atoms in the benzene ring on the canvas Then add the Fluor F atom The fluorobenzene fragment is now completed C aaa rs aleje SOLOA OJ SAA SESS E Jd eps alezo To start the search click the Substructure search button un
4. from MCI Koc from Kow L kg Log Koc from Kow Mass Amount Half Life Emissions kg hr hr Sewage Treatment Plant STP overall chemical mass balance using 10 000 hr EPI MPBPWIN v1 43 EPI HENRY WI N v3 20 EPI WATERNT v1 01 ACD ToxSuite 2 95 1 Ionization A CD Labs pKa ACD ToxSuite 2 95 1 Ionization A CD Labs LogD EPI KOAWIN v1 10 EPI AEROWIN v1 00 EPI KOCWIN v2 00 EPI Level III Fugacity Model EPI Suite v4 11 EPI STPWIN EPI Suite v4 11 20 Half Life d Half Life hr Overall Rate Const OH E 12 cm3 molecule sec and OZ E 17 cm3 molecule sec Biowinl linear model Probability of Rapid Biodegradation Biowin2 non linear model Probability of Rapid Biodegradation Biowin3 Expert Survey Ultimate Biodegradation Biowin3 Expert Survey Ultimate Timeframe Biowin4 Expert Survey Primary Biodegradation Biowin4 Exp Survey Primary Timeframe Biowin5 MITI linear model Biodegradation Probability Biowin6 MITI non linear model Biodegradation Probability Biowin7 Anaerobic Linear Biodegradation Probability Petroleum Hydrocarbon Biodegradation Half Life days BCE L kg wet wt Log BCF L kg wet wt Whole Body Primary Biotransformation Fish Half Life days BCF Arnot Gobas upper trophic Including Biotransformation L kg wet wt BCF Arnot Gobas upper trophic Zero Biotransformation L kg wet wt BAF Arnot Gobas upper trophic Including Biotrans
5. lt 500 g mole Click the PhysChem button to the left on the main search screen and then select Mol WT g mole from the drop down menu Search condition Mol WT g mole lt 500 SABER ECT Full database Search Cancel Type in 500 in the blank field in the box and click the Search button The substances of interest will appear in the results window Example 2 Simple model endpoint search Search for molecules that are positive for Ames test Click the Human health button to the left on the main search screen and then select Genotoxicity Ames test Bacterial Reverse Mutation Test from the drop down menu The screen should now look like this Bacterial Reverse Mutation Test Ames test in S typhimuri Information Bacterial Reverse Mutation Test Ames test in S typhimurium in vitro Select predictions or experimental results Battery combines all three systems CASE Ultra Leadscope SciQSAR Experimenta from training set and search for structures predicted or experimentally tested Positive in applicability domain Negative in applicability domain Cancel The Ames test model is made in each of the three systems CASE Ultra Leadscope and SCIOSAR and in this example we will select the overall battery prediction Once the Battery is selected click the Positive button to start searching for molecules that are positive for Ames test The results will appear on the scre
6. pig and human Respiratory sensitisation in humans Esto e dg re 481 Sens 73 1 Spec 86 6 Conc 80 7 activation human in vitro Sci1QSAR Sens 77 9 Spec 80 8 Conc 79 6 CASE Ultra Sens 57 4 Spec 87 2 Conc 78 3 di 874 Sens 51 7 Spec 91 2 Conc 80 4 antagonism human in vitro Sci1QSAR Sens 56 3 Spec 91 1 Conc 81 9 CASE Ultra Q 0 59 Thyroid receptor a binding R 0 83 Q 0 68 log ICs9 in uM human in 118 Leadscope vitro i SciQSAR R 20 64 Q 20 57 2_ CASE Ultra DI Thyroid receptor p binding JE JE vitro SciQSAR R 0 65 Q 0 58 CASE Ultra Sens 72 4 Spec 83 9 Conc 78 5 a I 1684 Sens 80 4 Spec 80 4 Conc 80 4 human in vitro Sci1QSAR Sens 79 9 Spec 82 7 Conc 8 1 4 CASE Ultra Sens 65 0 Spec 85 1 Conc 76 4 a potential in 323 Sens 72 0 Spec 85 5 Conc 80 1 Humans Sci1QSAR Sens 64 6 Spec 92 7 Conc 81 4 CASE Ultra Sens 89 7 Spec 95 1 Conc 91 9 Ashby structural alerts 782 Sens 87 5 Spec 90 7 Conc 88 5 Sci1QSAR Sens 81 7 Spec 80 6 Conc 8 1 1 17 CASE Ultra Sens 83 9 Spec 89 1 Conc 86 4 4102 Sens 84 3 Spec 85 7 Conc 84 9 Sci1QSAR Sens 79 3 Spec 79 1 Conc 79 2 Direct acting Ames CASE Ultra Sens 63 5 Spec 90 4 Conc 79 5 ONLY use for Ames Leadscope Sens 66 9 Spec 78 9 Cone 74 0 ONT ise for Ames 388 Leadscope Sens 66 9 Spec 78 9 Conc 74 0 POS_IN Sci1QSAR Sens 56 5 Spec 72 9 Conc 68 6 CASE Ultra Sens 52 8 Spec 88 4 Conc 71 9 Base pair Ames mutagens 204 ONLY use f
7. selected individually The battery prediction approach is further described below in the section Battery algorithm It is also possible to select and search for experimental results from the training set Searches Results Substances Bacterial Reverse Mutation Test Ames test in 5 typhimura a Information Structure Bacterial Reverse Mutation Test Ames test in typhimurium in vitroj Select predictions or experimental results PhysChem Battery combines all three systems ADME CASE Ultra Leadscope Environment Fa D Y ES SciQSAR Experimental from training set and search for structures predicted or experimentally tested Positive Intersect results in applicability domain OR Unite results Hegative in applicability domain NOT LES Complement results Clear Figure 6 Dialog box from query shown in Figure 5 Select the relevant results type predictions experimental in the dialog box and then click either the positive or negative button to start the search Only the predictions within applicability domain will be searched and displayed 10 The search will generate the browser window shown in Figure 8 and give the possibility to download a report containing the prediction results in the rtf file format The results window is further described below in the section Results window with substances Information about the selected model can be found by selecting the information
8. tab at the top of the dialog box A list of options will appear enabling you to download QMRFs of the relevant model versions Search examples are given in Appendix 3 Combining searches Combinations of searches are also possible These are performed using the two buttons to the left on the main screen AND and OR Using the OR button will display all substances from two or more searches whereas the AND button will only display the intersection of the individual searches The individual queries are made as described in the previous text so that they appear under Searches on the main search screen To combine searches click the search definition buttons for the searches of interest that appear under the field Searches This will highlight the text in the selected buttons which change color to green Then select either the AND or OR button to start the search The results of the combination search are displayed to the right under Searches Results and Substances The example in figure 7 shows the result of a combination of searches for AR antagonism and PXR binding using the AND button The result window is further described below in the section Results window with substances The NOT button to the left on the main search screen 1s for inverting a search Click the search definition button of interest only one under the field Searches and select NOT Inverted searches as well as results of AND and OR searc
9. Conc 71 3 CASE Ultra Sens 31 1 Spec 92 0 Conc 70 9 PEP EU MEM ET Sens 35 6 Spec 88 6 Conc 69 3 or mouse in vivo Sci1QSAR Sens 38 5 Spec 84 8 Conc 69 1 Sens sensitivity Spec specificity Conc concordance Ext validation external validation RI reliability index N CO ON FDA RCA cancer male rat in vivo FDA RCA cancer female rat in vivo FDA RCA cancer rat in VIVO FDA RCA cancer male mouse in vivo FDA RCA cancer female mouse in vivo FDA RCA cancer mouse in VIVO FDA RCA cancer rodent in VIVO Table 3 Software names and versions for physical chemical and environmental models Predicted property Melting Point deg C Boiling Point deg C Melting Point Experimental deg C Boiling Point Experimental deg C Vapour Pressure mm Hg Vapour Pressure Pa Vapour Pressure Experimental mm Hg Vapour pressure Subcooled Liquid Pa HLC Bond Method atm m3 mole HLC Group Method atm m3 mole HLC Via VP WSol atm m3 mole HLC Via VP WSol Pa m3 mole Henrys Law Const Exp db Pa m3 mole Henrys Law Const Exp atm m3 mole Water solubility from Kow mg L Water solubility Exp mg L Water solubility Exp Ref Log Kow Log Kow Exp Log Kow Exp Ref Water solubility from Fragments mg L pKa Acid pKa Base Log Koa Log Kaw Kp m3 ug Mackay based Kp m3 ug Koa based Phi Junge Pankow based Phi Mackay based Phi Koa based Koc from MCI L kg Log Koc
10. D list search type or paste in the query in the white box and click the Search button Choosing Affiliation gives two options for retrieving database structures REACH Pre registration list and PubChem To retrieve the structures choose the database of interest and click the Search button Search example is given in Appendix 1 Searching by structure The Structure section offers a 2D fragment editor where it is possible to build structure fragments to search for Click the Structure button to the left on the main search screen to open the edit box see Figure 4 Structures can be drawn and searched for by using the tools in the box It also contains other functionalities such as SMILES pasting and name lookup Mew search Substructure Similarity substructure search Exact match search JOANA AP E E E a E x 7 7 EA 5 Intersect resulte OR Unite results NOT Cancel Complement results Clear Figure 4 The Structure search interface Building a fragment To add an item click on the corresponding button and then click on the blank canvas Add atoms fragments bonds one by one To start a structure search select either Substructure or Similarity to the right on the screen When choosing Substructure two options are possible Select Substructure search to search for the built fragment as a subfragment within the database or Exact match search to search for the exact same structure When choosing Si
11. H The results in the database have also been used to generate the Danish Advisory Self classification List and to screen for potential PBTs Besides direct replacement of experimental tests in some cases QSAR predictions can help prioritize further in vitro and in vivo testing of chemicals In cases where animal testing is still needed QSAR predictions of mechanistic properties for the chemical can contribute in optimizing the experimental design In this way QSARs can reduce the need for later animal testing It is anticipated that the use of QSAR predictions and hence the need for good tools will grow in the future The new version of the QSAR database has been rebuilt from scratch and 1s an updated extended and improved version of the previous 2004 version of the online QSAR predictions database It contains an improved user friendly interface new functionalities and updated predictions for a considerably larger substance structure set than the previous database The new database is a dynamic system which will be updated continuously in terms of functionalities and content Introduction The new Danish QSAR database is a repository of model estimates for more than 600 000 substances The QSAR models include endpoints for physico chemical properties environmental fate bioaccumulation eco toxicity absorption metabolism and toxicity As far as possible all organic single constituent substances that were pre registered under REACH around
12. Sens 87 3 Speo 85 2 Conc 864 POS Not Ready 735 Leadscope Sens 87 3 Spec 85 2 Conc 86 4 Sci1QSAR Sens 63 0 Spec 92 7 Conc 77 8 CASE Ultra No robust model 565 Fathead minnow 96h LC50 mg L eere magna 48h EC50 626 R2 0 67 Q 0 64 531 Pseudokirchneriella s 72h EC50 mg L SciQSAR R 0 64 Q 0 60 CASE Ultra Sens 43 9 Spec 87 0 Conc 74 1 Cytochrome P450 2D6 CYP2D6 substrates 746 Sens 60 0 Spec 89 4 Conc 80 1 human clinical data Sci1QSAR Sens 59 5 Spec 79 8 Conc 73 1 CASE Ultra Sens 30 6 Spec 83 6 Conc 68 8 Cytochrome P450 2C9 CYP2C9 substrates 736 Sens 30 0 Spec 89 6 Conc 75 4 human clinical data Sci1QSAR Sens 26 3 Spec 91 5 Conc 74 7 T ACDLabs Ext validation RISOS Q 0 6 z M 375 ACDLabs Ext validation RI gt 0 5 Q 0 56 2 14 678 ACDLabs Ext validation RI gt 0 5 Q 20 55 OF 3 27 004 ACDLabs Ext validation RI gt 0 5 Q 0 61 2 14 972 ACDLabs Ext validation RI gt 0 5 Q 0 66 2 6432 ACDLabs Ext validation RI gt 0 5 Q 0 57 CASE Ultra Sens 69 4 Spec 92 5 Conc 82 5 Maximum recommended daily dose MRDD in 1222 Sens 78 6 Spec 82 5 Conc 80 7 humans lt 2 69 mg kg 2bw d Severe skin irritation in Sens 79 5 Spec 81 7 Conc 80 6 Sci1QSAR Sens 77 3 Spec 71 3 Conc 74 3 rabbit Sci1QSAR Sens 73 1 Spec 77 3 Conc 75 3 CASE Ultra Sens 63 4 Spec 86 7 Conc 75 8 836 79 5 81 16 o age Raper an Allergic contact dermatitis in guinea
13. User Manual for the Danish O SAR Database 23 November 2015 Copyright notice terms and conditions of use Permission is granted to use information from the database as is The database is an expert tool where the final assessment of properties is not dictated by the Q SAR estimates but by the user s own scientific judgment Aside from the fact that models are never perfect the Q SAR field is under rapid development and models are regularly updated and improved It is also impossible to provide the detailed information accompanying each individual prediction that 1s available to those who do not own licences to the software platforms The structural information in the database stems from many sources and in some cases 1t may be wrong The structures are also in some cases abbreviated in that possible anions and cations have been removed This can have important toxicological significance e g for Heavy Metal salts All access to the database should happen through the provided client side software and without any use of automated workflow or scripting Reproduction of information from the database is permitted provided the source is acknowledged as follows Danish Q SAR Database Division of Diet Disease Prevention and Toxicology National Food Institute Technical University of Denmark http qsar food dtu dk The Technical University of Denmark DTU is not responsible for any errors or inaccuracies the database may contain and i
14. car Hne WY STUC A a m 8 Seac hine Dy model endo ood aru cte AM du MIU ELM MI M DEA M AMD 9 Combine Sede De So nia 11 Searches and Re sulls Sec LIONS a d 12 Results Window WII SUD LANG CS ia sine hecdatacntelieniueneltune uod O ibas tO tunt iN med ss E 12 Technicalqequirements and nota add daa 14 battery ale OBI anita iia 14 Appendix 1 Searching by identification number ooooonnnnnnnncnnnnnononnnnnnnnnnnnonannnnnnnnnnnonnnannnnnnnnnnnnnnnos 21 Appendix 2 Searching by structure and similarity lessen 22 Appendix 2 Searching by model endpomMti ra ici 26 Appendix 4 Combining Searches mess dll dicta 27 Appendix 5 Software systems used for ModellNg ooooonccnnnnccnnnnnononcnnnnnnnnnnncnnnnnnnnnnonnnnnnnncnnnnnnnnnnnos 28 Background The Danish QSAR database has been freely available on the internet since 2004 It is a tool that allows industry research authorities and others to search for hazard information on chemical substances especially those with little or no testing data The information provided may be useful to identify chemical substances of potential concern With the EU chemicals legislations e g the REACH regulation there is increased focus on the use of alternatives to animal testing The QSAR database is used for a wide variety of tasks such as screening for potentially harmful substances and for assessment of specific substances e g in relation to dossier evaluation under REAC
15. cell transformation Sens 71 6 Spec 76 5 Conc 74 5 in vitro SciQSAR Sens 76 1 Spec 66 5 Conc 71 3 18 CASE Ultra Sens 75 4 Spec 92 0 Conc 83 6 Sex linked recessive lethal SLRL test in Drosophila m 367 Sens 79 1 Spec 80 3 Conc 79 6 in vivo Sci1QSAR Sens 74 2 Spec 78 3 Conc 76 2 CASE Ultra Sens 31 2 Spec 95 2 Conc 75 7 pone ok E 357 Sens 64 1 Spec 77 6 Conc 72 3 erythrocytes in vivo Sci1QSAR Sens 52 1 Spec 83 3 Conc 69 7 CASE Ultra Sens 42 4 Spec 92 7 Conc 73 7 o LT Sens 61 5 Spec 80 4 Conc 71 8 rodents in vivo Sci1QSAR Sens 57 7 Spec 81 4 Conc 71 7 CASE Ultra Sens 91 8 Spec 94 8 Conc 93 9 Sister chromatid exchange in mouse bone marrow cells in 265 Sens 88 6 Spec 95 9 Conc 94 0 VIVO Sci1QSAR Sens 76 7 Spec 93 2 Conc 86 8 CASE Ultra Sens 60 1 Spec 93 1 Conc 82 9 Sus Sens 86 6 Spec 80 8 Conc 83 1 Sci1QSAR Sens 82 4 Spec 82 0 Conc 82 2 CASE Ultra Sens 34 2 Spec 95 0 Conc 63 9 1324 Sens 62 6 Spec 74 7 Conc 69 2 CASE Ultra Sens 44 4 Spec 93 3 Conc 71 6 1321 Sens 57 7 Spec 83 6 Conc 72 7 CASE Ultra Sens 41 7 Spec 94 0 Conc 66 9 1379 Sens 57 1 Spec 82 3 Conc 71 2 CASE Ultra Sens 38 4 Spec 86 1 Conc 66 1197 Sens 58 6 Spec 81 4 Conc 71 9 CASE Ultra Sens 41 5 Spec 85 9 Conc 65 6 1208 Sens 59 2 Spec 80 6 Conc 71 3 CASE Ultra Sens 43 1 Spec 86 9 Conc 66 9 1221 Sens 56 5 Spec 83 9 Conc 72 7 CASE Ultra Sens 51 4 Spec 88 3 Conc 68 2 1530 Sens 65 9 Spec 76 2
16. der the Substructure heading to the right in the edit box The molecules that contain the fluorobenzene fragment will be shown to the right on the screen Substances Previous next a 2f 3 Lanza Structures 1 10 of 40734 m Structure Id The total number of molecules containing the fluorobenzene fragment are shown under the heading Results The functionalities of the results window are described in detail in the section Results window with substances in the manual Example 2 Exact match search Find molecules exactly matching the fluorobenzene structure Start by drawing the fluorobenzene structure as described in example 1 To start the search click the Exact match search button under the Substructure heading to the right in the edit box You will now see the search result in the window under Substances to the right on the screen 23 Example 3 Similarity search Find the most similar chemicals to fluorobenzene Start by drawing the fluorobenzene structure as described in example 1 Then click the heading Similarity to the right in the edit box Before you start the search you need to select if all structures or a user defined number e g 100 should be displayed In this example the 100 closest analogs is selected Am ejns B ERIGI Similarity OJOJDI v OI J SJ AA MASSAS JE els ol ado Cancel To start the search click the Similarity button The resulti
17. en 26 Appendix 4 Combining searches Example 1 Complex search containing fragments and model endpoints Search for molecules that have a fluorobenzene fragment molecular weight lt 500 g mole a positive Ames test We have already made searches for each of the three criteria in the previous examples When the three individual searches are made one by one the screen should look like this Searches Results 1 Substructure mu rA Mol WT g moie 500 nin ELE 3 POS Battery Bacterial Reverse Mutation Test Ames test 31813 To search for molecules that meet all three criteria left click on the three search strings This will change the color to green Now click on the AND button to the right on the main search screen to start the search At this point you should see this on the screen Searches Results The molecules that meet all three criteria will be shown in the window under Substances to the right on the screen Substances 1 AND 2 AND 3 Page 1 Previous next LaL3 La Structures 1 10 of 1102 Structure Id Similarity Appendix 5 Software systems used for modeling Case Ultra CASE Ultra divides each substance into fragments containing 2 10 interconnected atoms non hydrogen atoms These fragments are labelled with the experimental value of the parent substance as active or inactive If a fragment is over represented p gt 95 in the group of active or inactive sub
18. formation L kg wet wt BAF Arnot Gobas upper trophic Zero Biotransformation L kg wet wt LC50 Fish or EC50 Daphnid and Algae for Most Toxic Class mg L Max Log Kow for Most Toxic Class Most Toxic Class Lipinski s Rule of five bioavailability Absortion from gastrointestinal tract for 1 mg dose 96 p Absortion from gastrointestinal tract for 1000 mg dose 96 Log brain blood partition coefficient Dermal absorption mg cm2 event Acute toxicity in rodents Rat Oral Rat Intraperitoneal Mouse Oral Mouse Intraperitoneal Mouse Intravenous Mouse Subcutaneous Predicted property EPI AOPWIN v1 92 EPI BIOWIN v4 10 EPI BCFBAF v3 01 EPI ECOSAR v1 11 Equations from literature EPI DERMWIN v2 02 ACD ToxSuite 2 95 1 Appendix 1 Searching by identification number Example 1 Single ID query Search for registry number 80 05 7 Start by clicking the Id button to the right on the main search screen The ID Search box will appear on the screen ID Search Single ID ID List Affiliation Registry Number Q EC Number O PubChem CID Q Chemical name E Search E Cancel Select Single ID and Registry number from the list in the ID Search box Once you ve done this type in the registry number with or without hyphens in the blank field To start the search click the Search button You will now see the result to the right on the screen under Substances Substances RN 80 05 7
19. gularly updated and improved It is also impossible to provide the detailed information accompanying each individual prediction that is available to Human health those who do not own licences to the software platforms The structural information in the database stems from many sources and in some cases it may be wrong The structures are also in some cases abbreviated in that possible anions and cations have been removed This can have important toxicological significance e g for Heavy Metal salts AND All access to the database should happen through the provided client side software and without any use of automated workflow or scripting OR Ms The Technical University of Denmark is not responsible for any errors or a inaccuracies the database may contain and is not liable for any use that may be NOT made of the information contained therein Complement results Reproduction of information from the database is permitted provided the source is acknowledged as follows Danish Q SAR Database Division of Diet Disease Prevention and Toxicology National Food Institute Technical University of Denmark http qsar food dtu dk Clear Figure 2 Main search screen with disclaimer box Main search screen In the left part of this screen a number of buttons and the headline New Search is shown There are three basic search options in the interface window ID Structure and Model endpoint divided
20. hes can in turn be combined with other individual or combined ones to form more complex combined searches Search example is given in Appendix 4 11 New search Searches Results Substances 2 POS Battery Androgen Receptor AR antagonism human in 2 8100 AND 2 3 Page 1 id 3 POS Battery Pregnane X Receptor PXR Binding human in 2 69439 Previous Next 1 2 3 4 AND 2 3 1997 Structures 1 10 of 1997 Structure Similarity PhysChem ADME Environment Human health AND Intersect results OR Unite results NOT Complement results Clear Figure 7 An example where searches for AR antagonism and PXR binding are combined by using the AND button Searches and Results sections Every time you perform a search several new screen elements will appear A search definition button will be added to the Searches section Figure 7 It can be used for combining searches which is described in the section Combining searches Another button in the Results section will display the number of structures resulting from the search The actual structures will be listed in a browser window similar to the window shown in Figure 8 described below in the section Results window with substances The Searches and Results sections will keep track of all searches you have performed You can clear
21. into PhysChem ADME Environment and Human health These are explained in more detail below Each search can be combined with others in order to form more complex search queries The combined searches are performed using the three buttons AND OR and NOT and are described in more detail below in the section Combining searches The Clear button is used to clear the previous searches from the screen Searching by identification data The ID search button is designed for queries by Single ID ID List or Affiliation see Figure 3 When choosing Single ID a number of options are possible Registry Number EC Number PubChem CID and Chemical name To start a Single ID search type in the query in the white box and click the Search button The Registry number can be typed both with or without hyphens The structures matching your search will be listed in a browser window similar to the window shown in Figure 8 and give the possibility to download a report containing the prediction results of the resulting substance The search section and the result window are further described below in the sections Searches and Results and Results window with substances New search Searches Results Substances ID Search e a Registry Number EC Number PubChem CD Chemical name Search Figure 3 The ID search box When choosing ID list in the ID Search box two options are possible Registry numbers and PubChem CIDs To start an I
22. milarity either all analogs in the database or a user defined number of closest analogs will be displayed The search will generate a browser window similar to the window shown in Figure 8 and give the possibility to download a report containing the prediction results The results window is further described below in the section Results window with substances Editor operations Undo Undoes the last operation Redo Repeats the last undone operation Center Moves the fragment to the center of the canvas Toggle R S labels Marks R S isomeri Clear Clears the Edit window Import Imports MOL or SMILES file Export Displays MOL or SMILES information for current structure About Displays version number etc of the fragment editor 2D cleanup depiction Corrects bond angles etc Chemical dictionary search Name lookup in PubChem dictionary Search examples are given in Appendix 2 Searching by model endpoint The PhysChem Environment ADME and Human health buttons to the left on the main screen can be used to search for specific model endpoints Each of the four categories covers a number of different endpoints To start a search by model endpoint click the category button of interest e g Human health This will generate a drop down menu with a list of subcategories as shown in Figure 5 Figure 5 shows an example query to search for prediction results in the model for Bacterial Reverse Mutation Test Ames test in S typhimurium i
23. more than 10 structures they are shown in pages with 10 structures per page Use the top button row Previous Next First Current Last etc to navigation through the result pages All result pages are directly accessible the moment the search is executed so you can e g view any page directly without having to go first through the preceding ones KA substance of interest This will provide an RTF file containing all predictions as well as training set data when available The RTF document format is supported by Microsoft Word OpenOffice and other viewers editors To download a single substance report click the button in the id column next to the Clicking the Similarity button will open the 2D fragment editor where it is possible to search for substances similar to a query substance within the current result set The current result set will be ordered by decreasing similarity to the query substance To revert back to the Id order click the Id button above the structure list Clicking the button next to Similarity opens a dialog box where you can select any database property experimental or predicted in any model and predictive system and display its values in the result window You can select up to eight properties to display The extra information will be displayed in new columns and refresh as you navigate through result pages The Substances window can be resized and moved and scroll bars will automatically appear if nece
24. n vitro which is found in the genotoxicity subcategory New search Searches Results Substances Structure PhysChem ADME Environment Human health Acute toxicity Irritation and Sensitisation b Endocrine and Molecular Endpoints Developmental Toxicity b Genotoxicity Ashby Structural Alerts for DNA Reactivity Ames test Bacterial Reverse Mutation Test Ames test in S typhimurium in vitro Carcinogenicity in vivo Other in vitro endpoints Direct Acting Ames Mutagens without 59 ONLY use for Ames POS_IN In vive endpoints Base Pair Ames Mutagens ONLY use for Ames POS_IN Frameshift Ames Mutagens ONLY use for Ames POS_IN Potent Ames Mutagens Reversions gt 10 Times Controls ONLY use for Ames POS_IN Figure 5 An example query to search for prediction results in Ames test As shown a number of submodels are available When the model of interest 1s chosen a dialog box appears Figure 6 Select the heading Search at the top of the dialog box to start a search The menu in the dialog box depends on whether the selected model is made in one or more software systems The selected model in Figure 6 is made in three systems CASE Ultra Leadscope and SciQSAR Based on predictions from the three systems a fourth and overall battery prediction is made These four predictions three predictions from the individual systems and the battery prediction can be
25. ng substances will be ordered by similarity to the query chemical Example 4 Similarity search Find the REACH chemicals that are most similar to fluorobenzene Start by searching the REACH chemicals in the database Click the D button to the left on the main search screen and then click the heading Affiliation in the ID search box The box should now look like this Affiliation Search PTS Single ID ID List Affiliation Retrieve all database structures with affiliation in amp REACH Pre registration list Q Pubchem T Search Select REACH pre registration list and then click Search This will generate a results window showing all REACH pre registered chemicals 24 REACH Pre registration list Page 1 Structures 1 10 of 72524 gt Now you can search for similarity within the REACH pre registration list Click the Similarity button marked with a red arrow in the results window above At this point a new box will open anar esos The current result set 72524 structures will be ordered by similarity to the query chemical Similarity NODO eee KIHOLELA Cancel Draw the fluorobenzene structure described in example 1 on the blank canvas and click the Similarity button to start the search The REACH substances will now be ordered by similarity to the query chemical 25 Appendix 3 Searching by model endpoint Example 1 Simple model endpoint search Search for molecules with molecular weight
26. or Ames Sens 70 2 Spec 66 4 Conc 68 4 POS_IN Sci1QSAR Sens 68 6 Spec 67 7 Conc 68 1 CASE Ultra Sens 73 5 Spec 84 1 Conc 78 9 ONLY use for Ames 309 Sens 74 4 Spec 78 6 Conc 76 6 POS_IN Sci1QSAR Sens 68 3 Spec 78 2 Conc 73 8 Potent Ames mutagens CASE Ultra Sens 73 7 Spec 87 7 Conc 81 2 reversions gt 10 times E H E vonmolos ONLY uce tof 187 Sens 68 9 Spec 70 0 Conc 69 8 Ames POS_IN SciQSAR Sens 75 0 Spec 74 7 Conc 74 9 CASE Ultra Sens 40 4 Spec 94 5 Conc 74 4 Chromosome aberrations in COCR wii 233 Sens 54 1 Spec 79 3 Conc 68 8 SciQSAR Sens 50 5 Spec 84 3 Conc 70 3 CASE Ultra Sens 63 3 Spec 86 7 Conc 76 4 Sens 74 6 Spec 75 2 Conc 74 9 SciQSAR Sens 73 0 Spec 72 8 Conc 72 9 Mutations in thymidine CASE Ultra Sens 76 5 Spec 86 3 Conc 81 2 ase TOSAN Sens 85 1 Spec 83 8 Conc 84 4 lymphoma cells in vitro SciQSAR Sens 79 1 Spec 80 5 Conc 79 8 CASE Ultra Sens 75 4 Spec 84 5 Conc 78 9 Mutations in HGPRT locus in CHO cells Sens 81 7 Spec 78 4 Conc 80 5 in vitro SciQSAR Sens 80 0 Spec 73 0 Conc 76 5 CASE Ultra Sens 60 6 Spec 87 0 Conc 74 1 Bacterial reverse mutation test Ames test in S typhimurium in vitro Frame shift Ames mutagens Chromosome aberrations in CHL cells in vitro Unscheduled DNA synthesis UDS in rat hepatocytes in Sens 74 1 Spec 70 1 Conc 72 4 vitro SciQSAR Sens 69 6 Spec 72 5 Conc 71 1 CASE Ultra Sens 50 8 Spec 86 9 Conc 74 0 Syrian hamster embryo SHE
27. previous pilot project including 32 different models and the three systems mentioned above not published For a given effect QSAR predictions are made in each of the independent QSAR model systems and combined into a battery prediction by using the criteria shown in Table 1 The first column shows the total number of predictions positive negative in domain The next two columns show the number of positive and negative predictions respectively The final battery prediction based on the individual predictions 1s shown in the fourth column Table 1 Battery algorithm Total POS NEG POS NEG Battery prediction Remarks 3s 3 0 homn E 0 h mon 2 pon 1 INC_OUT EXCEPT when CU and LS are or see remark both NEG_IN in this case the NEG_IN battery call is NEG_IN 2 o POS AN IN INC_OUT 3 14 Hn i 9 foo IN o 06 mor OoOo INC_OUT If minimum one prediction out of domain O None predicted POS positive NEG negative INC inconclusive IN inside applicability domain OUT outside applicability domain Less weight is put on an SQ POS compared to LS or CU POS in cases where LS and CU agree on a NEG in AD prediction because SQ in many cases has lower specificity than LS and CU 15 Table 2 Training set numbers and cross validation results See QMRFs for more information Cross validation result CASE Ultra Sens 68 9 Spec 87 8 Conc 77 2 POS Not Ready Leadscope
28. ptors such as logP molecular weight and the number of hydrogen bond acceptors and donors When a model is developed a sub set of the structural features and the physico chemical descriptors are used Features and descriptors can be chosen either manually or automatically SciQSAR The software operates with various molecular descriptors e g physical chemical electrotopological state E state and hydrogen electrotopological HE state indices connectivity indices and other descriptors The E state is a value calculated for each atom or hydride group e g CH3 NH OH N and Cl in a molecule It is large for electronegative atoms especially for those with a few skeletal connections and is smaller for less electronegative atoms and atoms with several o bonds The HE state is calculated for each hydrogen connected to non hydrogen atoms e g OH NH NH CH CH3 in a molecule This index shows the polarity of the hydrogen where a highly polar hydride group gives a greater value e g OH than a less polar group e g CH3 The electrotopological state indexes E state and HE state are related to an attribute called electron accessibility This attribute comes from the hypothesis that atoms with higher electron density that are accessible for contact with other molecules have a higher potential for interaction The molecular connectivity descriptors x are calculated for the whole molecule This group of descriptors provides information on skeletal va
29. riation including degree of branching and types of branching 28
30. s not liable for any use that may be made of the information contained therein DTU do not warrant and hereby disclaim any warranties with respect to the accuracy adequacy or completeness of any information obtained from this database Nor do we warrant that the site will operate in an uninterrupted or error free manner or that the site and its components are free of viruses or other harmful components Use of information obtained from or through this site is at your own risk As a user of this database you agree to indemnify and hold DTU harmless from any claims losses or damages including legal fees resulting from your use of this database and to fully cooperate in DTU s defense against any such claims These terms are governed by Danish Law with the exception of international private law and conflict of law rules to the extent that such rules would result in the application of another country s law Any dispute arising between the parties in connection with the use of this database including the interpretation of the above terms which cannot be settled amicably by negotiation between the parties shall be settled by the Court of Lyngby Denmark as the court of first instance Contents Copyrebtnotice terms and conditions OT US iaa 2 BAR MN NER cence cet cra TN ER P 4 to duce 4 Maine atures ata placas 5 Launching the Danish OS AR Dita E AA 5 Man AIC O a a a T T amet 6 Searc hine DY THE DB HOO OT data a a E a a a Ras dep S 6 S
31. ssary 13 Technical requirements and notes All operations with the Danish Q SAR database are performed in a web browser There is no need to download or install any software Likewise there is no need to install any browser plugin or add on the previous version of the web site used Java The system is can be accessed from both personal computers and mobile devices The minimum screen resolution for using the system 1s 640x480 pixels For convenience higher resolution display settings can be recommended preferably 1280 or more pixels on the horizontal axis The client side software 1s implemented entirely in JavaScript and is compatible with all major browsers and operating systems without the need for third party software Depending on the security settings of your browser you may need to enable JavaScript in order to use the website The system has been tested with the following browser versions Google Chrome 46 0 Microsoft Internet Explorer 11 Opera 33 0 Mozilla Firefox 37 0 2 Battery algorithm Some of the models are made in two or three of the following independent systems CASE Ultra CU Leadscope Predictive Data Miner LS and SciQSAR SQ The systems are described in Appendix 5 Based on predictions from each of the applied systems a battery prediction is made using a so called battery algorithm The battery approach can give more reliable predictions and can also expand the applicability domain which was shown in a
32. stances the fragment is assumed to be relevant for the modelled activity If the fragment is not significantly overrepresented in active or inactive substances it will not be considered important A fragment with a statistical correlation to active or inactive substances is called a biophore or a biophobe respectively When all fragments have been examined for their importance to activity a hierarchical selection takes place starting with the biophore with the best statistical significant result Substances containing this substructure are set aside and the next biophore is found in the same manner This is repeated until either the entire training set is used or there are no more statistically significant fragments The whole procedure is then performed for biophobes in an identical way Each group of substances containing a biophore or biophobe is then analysed to find modulators that either enhance or decrease the probability of the fragment being a biophore biophobe The modulators can be structural fragments or chemical properties e g activating fragments deactivating fragments logK molecular orbital energies Leadscope When data are imported into Leadscope the substances are classified by structure into categories using a library of approximately 27 000 structural features The structural features are substructures such as functional groups heterocycles and pharmacophores The program also calculates a number of physico chemical descri
33. them by clicking the Clear button or delete individual searches using the small gt button next to the search definition You can revisit previous search results at any time by clicking the button displaying the number of structures in the Results section The relevant searches are not executed again but instead retrieved quickly from a repository of searches Results window with substances The searches described in the previous sections will generate a results window and give the possibility to download a report containing database results for selected substances One report per substance will be generated The example in Figure 8 shows the result of a Model endpoint search of the Bacterial Reverse Mutation Test The results of the search are displayed to the right under Results and Substances 12 New search Searches Results Substances i POS Battery Bacterial Reverse Mutation Test Ames test 2 31813 POS Battery Bacterial Reverse Mutation Test Ames test Page 1 Previous Next 1 23 3182 Structures 1 10 of 31813 Structure PhysChem ADME Human health AND Intersect results OR Unite results NOT Complement results Figure 8 An example of a results window from a Model endpoint search of the Bacterial Reverse Mutation Test The window under Substances shows the resulting structures When there are
Download Pdf Manuals
Related Search