DrugBank 4.0: shedding new light on drug metabolism
-
Upload
independent -
Category
Documents
-
view
0 -
download
0
Transcript of DrugBank 4.0: shedding new light on drug metabolism
DrugBank 40 shedding new light on drugmetabolismVivian Law1 Craig Knox1 Yannick Djoumbou2 Tim Jewison1 An Chi Guo1 Yifeng Liu1
Adam Maciejewski1 David Arndt1 Michael Wilson1 Vanessa Neveu1 Alexandra Tang3
Geraldine Gabriel3 Carol Ly3 Sakina Adamjee3 Zerihun T Dame2 Beomsoo Han1
You Zhou1 and David S Wishart1234
1Department of Computing Science University of Alberta Edmonton AB Canada T6G 2E8 2DepartmentBiological Sciences University of Alberta Edmonton AB Canada T6G 2E8 3Faculty of Pharmacy andPharmaceutical Sciences University of Alberta Edmonton AB Canada T6G 2N8 and 4National Institutefor Nanotechnology 11421 Saskatchewan Drive Edmonton AB Canada T6G 2M9
Received September 15 2013 Revised October 12 2013 Accepted October 14 2013
ABSTRACT
DrugBank (httpwwwdrugbankca) is a compre-hensive online database containing extensive bio-chemical and pharmacological information aboutdrugs their mechanisms and their targets Since itwas first described in 2006 DrugBank has rapidlyevolved both in response to user requests and inresponse to changing trends in drug research anddevelopment Previous versions of DrugBank havebeen widely used to facilitate drug and in silico drugtarget discovery The latest update DrugBank 40has been further expanded to contain data on drugmetabolism absorption distribution metabolismexcretion and toxicity (ADMET) and other kinds ofquantitative structure activity relationships (QSAR)information These enhancements are intended tofacilitate research in xenobiotic metabolism (bothprediction and characterization) pharmacokineticspharmacodynamics and drug designdiscovery Forthis release gt1200 drug metabolites (including theirstructures names activity abundance and otherdetailed data) have been added along with gt1300drug metabolism reactions (including metabolizingenzymes and reaction types) and dozens of drugmetabolism pathways Another 30 predicted ormeasured ADMET parameters have been addedto each DrugCard bringing the average numberof quantitative ADMET values for Food andDrug Administration-approved drugs close to 40Referential nuclear magnetic resonance and MSspectra have been added for almost 400 drugs aswell as spectral and mass matching tools to
facilitate compound identification This expandedcollection of drug information is complemented bya number of new or improved search tools includingone that provides a simple analyses of drugndashtargetndashenzyme and ndashtransporter associations to provideinsight on drugndashdrug interactions
INTRODUCTION
DrugBank is a comprehensive repository of drug drugndashtarget and drug action information developed maintainedand enhanced by extensive literature surveys performed bydomain-specific experts and skilled biocurators Thequality breadth and uniqueness of its data have madeDrugBank particularly popular (gt8 million web hitsyear) and highly regarded among pharmaceutical re-searchers medicinal chemists clinicians educators andthe general public Because most of the data inDrugBank are expertly curated from primary literaturesources it has become the referential drug data sourcefor a number of well-known databases such asPharmGKB (1) ChEBI (2) KEGG (3) GeneCards (4)PDB (5) PubChem (6) UniProt (7) and Wikipedia Sinceits first release in 2006 DrugBank has been continuouslyevolving to meet the growing demands of its users and thechanging needs of its rapidly expanding user base Thefirst version of DrugBank was limited to providingdata on selected Food and Drug Administration(FDA)-approved drugs and their drug targets (8)Pharmacological pharmacogenomic and molecular biolo-gical data were added to DrugBank 20 along with a sig-nificant increase in the number of approved andexperimental drugs (9) DrugBank 30 released in 2010was expanded to include data on drugndashdrug and drugndashfood interactions metabolic enzymes and transporters as
To whom correspondence should be addressed Tel +1 780 492 0383 Fax +1 780 492 1071 Email davidwishartualbertaca
Nucleic Acids Research 2013 1ndash7doi101093nargkt1068
The Author(s) 2013 Published by Oxford University PressThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (httpcreativecommonsorglicensesby30) whichpermits unrestricted reuse distribution and reproduction in any medium provided the original work is properly cited
Nucleic Acids Research Advance Access published November 6 2013 at U
niversity of Alberta on N
ovember 25 2013
httpnaroxfordjournalsorgD
ownloaded from
well as pharmacokinetic and pharmacoeconomic informa-tion (10) For 2014 DrugBank has been enhanced tocapture the increasing body of quantitative knowledgeabout drugs and improved technologies to detect drugstheir metabolites and their downstream effects In particu-lar significant improvements and large-scale additions inthe areas of QSAR (quantitative structure activity rela-tionships) ADMET (absorption distribution metabol-ism excretion and toxicity) pharmacometabolomics andpharmacogenomics have been made Existing informationabout drug structures drug salt-forms drug names drugtargets and drug actions has also been expanded andupdated Numerous approved and experimental drugshave been added along with a number of new data fieldsdescribing each drug New search tools have also beendeveloped or improved on to increase the ease withwhich information can be foundMany of the enhancements made over the past 3 years
were stimulated by user feedback and suggestions We aregrateful to our users and continue to strive to meet theirneeds Further details on the additions and enhancementsmade to DrugBank 40 are described later
DATABASE ADDITIONS AND ENHANCEMENTS
The development and evolution of DrugBank includingprevious data content additions curation protocolsquality control methods general layout interfacefeatures and data sources has been described previously(8ndash10) Here we shall focus on enhancements and changesmade since the release of DrugBank 30 In particular wewill discuss (i) enhancements made to existing data (ii) theaddition of new data fields (iii) new and enhanced searchfeatures and (iv) a new user feedbackcommenting system
Enhancement of existing data
DrugBank has evolved significantly since its inception in2006 with a progressive expansion in the depth andbreadth of its data as well as improvements to thequality and reliability of its information The enhance-ments and additions in data content between DrugBank10 20 30 and 40 are summarized in Table 1 As seen inthis table DrugBank 40 has seen a further 40 increasein the number of data fields per DrugCard The number ofFDA-approved drugs has increased by 19 biotechdrugs by 14 and gt1200 new investigational drugs (inphase I II or III trials) have been added Likewise thenumber of drugs with transporters and carrier proteinshas increased by 40 and 30 respectively Additionallythe number of QSAR parameters has grown by 64 andthe number of ADMET parameters by 650 Thenumber of drugndashdrug interactions and the number ofdrugndashfood interactions have grown by 25 and 14respectively Similarly the number of drug target single-nucleotide polymorphisms (SNPs) has grown by 10whereas the number of SNPs associated with adversedrug effects (SNP-FX and SNP-ADR) has grown by78 In addition the number of illustrated drug actionpathways has grown by 38 Furthermore chemical syn-thesis information has been added to gt1200 drugs with
many synthesis references now hyperlinked to patents andreaction schemes maintained by Drugsynorg (httpwwwdrugsynorg)
Data quality and currency have always been two of thetop priorities for the DrugBank curation teamConsequently significant efforts over the past 3 yearshave been directed at improving the quality of existingDrugCards and keeping them as up-to-date as possibleUnlike DrugBank (ie the database) which has a 2ndash3-yearrelease cycle DrugBankrsquos DrugCards are continuouslyupdated as new information becomes available Becausenew drugs are constantly being approved and old drugsare perpetually being re-characterized re-purposed orremoved our curation team is continuously combing theliterature to update existing DrugCards or to add newDrugCards As a result hundreds of drug descriptionsdrug targets indications pharmacodynamics data mech-anisms of action and metabolism data fields have beenadded re-written andor expanded over the past 3 yearsWe have also updated and re-written all of DrugBankrsquosdrugndashdrug interactions In particular we have included orupdated all interactions where modifications in drugtherapy are recommended either by the manufacturer orother data sources
Drug nomenclature is particularly important forpharmacists physicians and the drug industry Manydrugs have dozens or even hundreds of alternate namesbrand names or synonyms often arising because ofvarious trademark patenting andor marketing require-ments For the latest release of DrugBank all of itssynonyms and brand names were extensively reviewedLengthy chemical and salt-form names were removedand only the most common synonyms have beenincluded enabling users to more easily browse alternatedrug names Brand names or synonyms appearing onmore than one DrugCard were also identified and editedor removed as required In total gt8000 duplicated or par-tially duplicated brand names and synonyms were editedor removed as part of DrugBank 40rsquos nomenclatureremediation effort
DrugBankrsquos experimental drug data set was alsoreviewed remediated and expanded In particular experi-mental drugs binding to nonndashhuman targets were removedwith the exception of those that bind to known bacterialviral and fungal pathogens Experimental drug targets ori-ginally referenced to the RCSB Protein Data Bank (5) arenow referenced to the original publications Additionalexperimental drugs were added from a variety of literaturesources and online databases such as ChEMBL (11)Further data quality checks on drug structures werecarried out by identifying InChI keys and ChEBI IDsappearing on more than one DrugCard In some casesduplicate IDs were the result of duplicate DrugCardswhereas in other cases the IDs referred to stereoisomersthat did not have stereochemistry depicted in their struc-tures These structures were manually corrected In total600 corrections of various forms were performed as partof the DrugBank 40 structure remediation effort
Drug classes are a useful way of grouping drugs withsimilar therapeutic effects and mechanisms of action Inthe latest version of DrugBank we expanded on the
2 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
information provided by the Anatomical TherapeuticChemical (ATC) Classification System defined by theWHO Collaborating Centre for Drug StatisticsMethodology (12) The first four levels of the classificationsystem (anatomical main group therapeutic subgrouppharmacological subgroup and chemical subgroup) arelisted for each drug with links to what we callCategoryCards The CategoryCard for each anatomicalmain group shows a listing of therapeutic subgroupswhich can be expanded to show pharmacological sub-groups which can be further expanded to show chemicalsubgroups which are expandable to finally show the fifthlevel chemical substances or drug names This new treestructure and the new browsing function called CategoryBrowse effectively replaces the old PharmaBrowse foundin earlier versions of DrugBank Adopting this well-estab-lished classification system in DrugBank 40 enables users tomore easily search for drugs with similar pharmacologiceffects
New data fields
As noted earlier significant efforts for this yearrsquos release ofDrugBank have been directed toward adding new data on
drug metabolism ADMET pharmacometabolomicspharmacogenomics and a variety of QSAR New datafields are also highlighted in Table 1 These additions toDrugBank 40 were motivated by the fast-moving devel-opments in the fields of xenobiotic metabolismpharmacometabolomics pharmacogenomics pre-clinicaldrug screening and computer-aided drug designWith regard to DrugBank 40rsquos effort to capture more
drug metabolism information it is important to note thatthe database now contains gt1200 drug metabolites(including their structures names activity abundanceand other detailed data) along with drug metabolismreaction data for gt1000 approved drugs These drug me-tabolism data include the sequence of reactions reactiontypes and whether drug metabolites are active inactiveandor major circulating entities In addition DrugBanknow offers comprehensive colorful manually drawninteractive drug metabolism pathways [using theSMPDB pathway model (13)] for gt50 drugs We believethat the addition of large numbers of drug metabolitesto DrugBank should also assist metabolomic andpharmacometabolomic studies by facilitating the identifi-cation of xenobiotics in biological samples (via MS
Table 1 Comparison between the coverage in DrugBank 10 20 30 and DrugBank 40
Category 10 20 30 40
Number of data fields per DrugCard 88 108 148 208Number of search types 8 12 16 18Number of illustrated drug-action pathways 0 0 168 232Number of drugs with metabolizing enzyme data 0 0 762 1037Number of drug metabolites with structuresa 0 0 0 1239Number of drug-metabolism reactionsa 0 0 0 1308Number of illustrated drug metabolism pathwaysa 0 0 0 53Number of drugs with drug transporter data 0 0 516 623Number of drugs with taxonomic classification informationa 0 0 0 6713Number of SNP-associated drug effects 0 0 113 201Number of drugs with patentpricingmanufacturer data 0 0 1208 1450Number of foodndashdrug interactions 0 714 1039 1180Number of drugndashdrug interactions 0 13 242 13 795 14 150Number of biochemical target-mediated drugndashdrug interactionsa 0 0 0 35 632Number of biochemical enzyme-mediated drugndashdrug interactionsa 0 0 0 226 463Number of biochemical transporter-mediated drugndashdrug interactionsa 0 0 0 60 644Number of biochemical drugndashdrug interactions (all)a 0 0 0 322 739Number of ADMET parameters (Caco-2 LogS) 0 276 890 6667Number of QSAR parameters per drug 5 6 14 23Number of drugs with drugndashtarget binding constant dataa 0 0 0 791Number of drugs with NMR spectraa 0 0 0 306Number of drugs with MS spectraa 0 0 0 384Number of drugs with chemical synthesis information 0 38 38 1285Number of FDA-approved small molecule drugs 841 1344 1424 1552Number of FDA small molecule drugs with salt-form structuresa 0 0 0 474Number of biotech drugs 113 123 132 284Number of nutraceutical drugs 61 69 82 87Number of withdrawn drugs 0 57 68 78Number of illicit drugs 0 188 189 190Number of experimental drugs 2894 3116 5210 6009Number of investigational drugs (phase I II and III trials)a 0 0 0 1219Total number of experimental and FDA small molecule drugs 3796 4774 6684 7561Total number of experimental and FDA drugs (all types) 3909 4897 6816 7713Number of all drug targets (unique) 2133 3037 4326 4115Number of approved-drug enzymescarriers (unique) 0 0 164 245Number of all drug enzymescarriers (unique) 0 0 169 253Number of external database links 12 18 31 33
aNew data
Nucleic Acids Research 2013 3
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
spectrometry) By providing the drug research communitywith gt1200 drug metabolites (along with their corres-ponding phase I and II enzymes) we are also hopingthat new and improved software could be developed forpredicting the structure of drug metabolites Efforts arealready underway in our laboratory to develop openaccess freeware for drug metabolism predictionIn addition to the substantial additions to DrugBankrsquos
drug metabolism data the quantity and diversity ofADMET data in DrugBank 40 has also been significantlyincreased Although a modest amount of ADMET datawas included in previous versions of DrugBank (8ndash10) itwas generally limited to a small number of approveddrugs However by taking advantage of several recentlyreleased software packages for predicting ADMETproperties (14) and mining newly published ADMETsurveys (1516) it has been possible to add 30 moreADMET data fields to nearly all of DrugBankrsquos drugentries These include measured andor predictedADMET values for Caco cell permeability (as a proxyfor intestinal permeability) bloodndashbrain barrier perme-ability human intestinal absorption levels P-glycoproteinactivity Cyp450 substrate preferences renal transportactivity carcinogenicity toxicity Ames test activity andhuman Ether a-go-go Related Gene (hERG) activityalong with their probability scoresAs the fields of metabolomics and pharmacome-
tabolomics continue to evolve we believe there is anincreasing need to develop drug databases that are com-patible with the needs of metabolomic researchers In par-ticular referential mass spectrometry (MS) and nuclearmagnetic resonance (NMR) spectra of pure drugs andtheir metabolites are particularly important for identifyingandor quantifying compounds in biological matricesPrescription and over-the-counter medications are fre-quently found in human biofluid samples characterizedin metabolomic studies To meet these compound identi-ficationcharacterization needs gt750 NMR and 3763 MSreference spectra for 400 different drugs have beenadded to DrugBank 40 Many of these spectra were ex-perimentally acquired by our laboratory (17) whereasothers have been assembled from freely available onlinesources (1819) In addition to including these referenceMS and NMR spectra we have also added a number ofadvanced spectral viewing and spectralmass matchingtools [originally developed for the HMDB (17)] to facili-tate metabolomic studies Using these tools users cansubmit chemical shift or mz (mass to charge ratio) listsand search against DrugBankrsquos spectral library for exactor even approximate matches Users can also browsezoom and view all of DrugBankrsquos spectral data tovisually compare results and assess spectral matches Tofurther enhance DrugBankrsquos compatibility withmetabolomic activities we have also added an extensivechemical taxonomy or chemical classification system thatuses the same hierarchical design and nomenclature asdeveloped for other metabolomic databases (1720) Inparticular compounds are classified into lsquoKingdomsrsquolsquoSuperclassesrsquo lsquoClassesrsquo lsquoSubclassesrsquo lsquoDirect ParentsrsquolsquoMolecular Frameworkrsquo lsquoSubstituentsrsquo and lsquoCross-Referencesrsquo using well-defined rules based on their
structural features and structure similarities This classifi-cation system allows users to identify all benzodiazepinesselect all alkaloids or extract all peptides in DrugBankusing a simple text query
To facilitate QSAR studies we have continued toadd to DrugBankrsquos collection of quantitative drug dataThis includes not only more molecular descriptors(number of H-bond donorsacceptors physiologicalcharge refractivity polarizability polar surface areaetc) but also more data on chemical substituents (seechemical taxonomy earlier) and extensive quantitative in-formation on drugndashligand binding constants The bindingconstant data have been mined from several sourcesincluding the primary literature and BindingDB (21)The combination of DrugBankrsquos existing 2D and 3Dstructural data along with new molecular descriptormolecular substituent and binding constant data for thou-sands of drugs should allow researchers to perform muchmore sophisticated and extensive QSAR studies
Many drugs exist in a variety of salt forms and thesesalt variants can play a significant role in the efficacydelivery or indication for certain drugs HistoricallyDrugBank has only provided the structure of lsquobasersquoform of the drug However to accommodate therequests of drug formulation experts and pharmacistswe have now added salt-form structures to DrugBank 40
New and enhanced search features
The quality of a database is determined not only by thequality of its content but also by the ease with which userscan access or navigate through its data In this regard theDrugBank team has continuously strived to improvethe databasersquos user interface to enhance its search utilitiesand to respond to user suggestions For DrugBank 40 wehave added two new browsing features to ease the retrievalof information (i) Reaction Browse and (ii) CategoryBrowse We have also improved the Multi-Searchfeature of DrugBankrsquos Interax (Drug InteractionLookup system) and streamlined the existing browsingutilities including DrugBrowse GenoBrowse andAssociation Browse A simple tool called BioInteractorfor rapidly retrieving and analyzing the biochemical datain DrugBank has also been added We have also upgradedDrugBankrsquos text search engine to improve the speed andquality of the search results and we have further enhancedDrugBankrsquos Data Extractor to make it easier to use andfar more accessible These new and improved searchbrowsing features should substantially increase the easewith which users are able to navigate retrieve andextract information from DrugBank
Reaction Browse was developed to facilitate the viewingand searching of drug metabolism reactions Its easilysearchable table format allows users to rapidly browsethrough pages displaying each of the drug metabolismreactions in DrugBank Each drug metabolite enzymeand reaction is hyperlinked to its respective DrugCardMetaboCard EnzymeCard and ReactionCard Searchbars above each column of the Reaction Browse tableenable users to search by drug name DrugBank ID me-tabolite name DrugBank metabolite ID enzyme name
4 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
UniProt ID or any combination of the above (Figure 1)Category Browse (which replaces the old PharmaBrowse)is a simple search tool that enables users to browse andsearch for drugs by category based on pharmacologicalaction or their ATC Classification (12) All four levels ofthe ATC classification system are listed together to facili-tate searching Each ATC drug category is hyperlinked toa CategoryCard as described earlier Furthermore eachdrug listed in the CategoryCard can be expanded toshow drug targets enabling users to easily compare drugtargets from the same category (Figure 1)
A new addition to DrugBankrsquos large selection ofsearch and browsing toolbox is called BioInteractorBioInteractor uses data on drug-target -enzyme and-transporter associations to provide insight on drugndashdrug interactions It allows users to identify forinstance which drugs act similarly on the same targetSuch drugs may have additive pharmacodynamic effectsif given concomitantly Conversely two drugs may antag-onize the beneficial effects of each other by havingopposite actions on a specific drug target Although clin-ically relevant drugndashdrug interactions are well docu-mented in commercial databases such as those producedby Lexi-Comp Inc the mechanisms of interactions arenot always clear BioInteractor uses ranked drugndashenzyme association information to predict enzyme-mediated drug interactions that may be clinicallyrelevant For this purpose enzyme inhibitors andinducers associated with DrugBankrsquos new drug metabol-ism data have been further categorized as lsquostrongrsquo lsquomod-eratersquo or lsquoneitherrsquo based on a set of criteria defined byBjornsson et al (22) Such a ranking system enablesusers to differentiate between drug effects on enzymeslikely to cause clinically significant changes to the pharma-codynamics of interacting drugs from those unlikelyto affect clinical outcomes A search query withBioInteractor will yield all possible enzyme-mediatedinteractions but the ranking system will provide furtherinsight on the probability of a clinically relevant inter-action occurring
In addition to these new search tools we have alsoimproved DrugBankrsquos standard search tools In particu-lar we have upgraded DrugBankrsquos text search engine touse Elasticsearch (httpwwwelasticsearchorg) anadvanced and scalable enterprise search backend It isalso now possible to search across drug targets drugndashdrug and drugndashfood interactions metabolic reactions me-tabolites and other relevant search fields In additionsearch results can now be downloaded using DrugBankrsquosupgraded Data Extractor The new Data Extractorappears in all of DrugBankrsquos standard browse andsearch windows allowing users to easily extract down-load and save their current data view at any point Inaddition all DrugBank data extractions performed byData Extractor are saved for up to 1 month so userscan return at any time to retrieve their results We havealso upgraded DrugBankrsquos Advanced Search thereby im-proving the granularity and coverage of data fields inDrugBank This should allow users to build much morecomplex queries through Advanced Searchrsquos simple inter-face DrugBankrsquos Advanced Search feature when
combined with the Data Extractor should allow users tosearch retrieve or extract almost any kind of data type orcombination of data within DrugBank
User feedback commenting system and news feed
Since the DrugBank was first released in 2006 the web hastransformed into a dynamic social environment Newsblogs and scientific online resources provide users withthe ability to leave comments and discuss articles Inresponse to these changes we have added a commentingfeature at the end of each DrugCard enabling users tosubmit comments that are moderated by the DrugBankcuration team This commenting system was designed toenable users to engage with the DrugBank curators toprovide a detailed record of changes and to stimulate con-versation among the scientific community Hundreds ofexcellent user comments have been received and actedon since this feature was implemented Additionally wehave created a DrugBank Twitter account allowingDrugBank curators to report news maintenance issuesand provide updates while at the same time allowingusers to supply feedback through social media Engagingour users has always been a priority and supplying avehicle to provide feedback will result in higher dataquality and an improved user experience
CONCLUSION
Over the past 7 years DrugBank has grown from a smallsomewhat specialized drug database to a large compre-hensive drug knowledgebase covering almost all aspects ofdrug function formulation and structure Throughout thisevolution we have strived to maintain the highest qualityand currency of data to continuously improve the qualityof its user interface and to attentively listen to all membersof its user community With this latest release ofDrugBank we have continued to make substantial im-provements to both the qualitycurrency of its data andthe quality of its user interface We have also added sub-stantial quantities of new and what we believe will behighly relevant data pertaining to drug metabolism drugmetabolites drug salt-forms ADMET and QSAR Theseadditions and enhancements are intended to facilitateresearch in xenobiotic metabolism (both prediction andcharacterization) pharmacometabolomics pharmacokin-etics pharmacodynamics pharmaceutics and drug designdiscovery They are also intended to make the data withinDrugBank relevant to a far wider audience including re-searchers in metabolomics pharmacogenomics andpharmacology formulation chemists pharmacists phys-icians educators and the general public Our commitmentto continue to maintain update and improve DrugBankwill continue for the foreseeable future Based on currenttrends in pharmaceutical and pharmacological research itis likely that future versions of DrugBank will includemany more drug action and drug metabolism pathwaysmuch more information regarding the transport anduptake of drugs as well as greater detail regarding themolecular mechanisms underlying drug toxicity anddrug-associated adverse reactions
Nucleic Acids Research 2013 5
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
well as pharmacokinetic and pharmacoeconomic informa-tion (10) For 2014 DrugBank has been enhanced tocapture the increasing body of quantitative knowledgeabout drugs and improved technologies to detect drugstheir metabolites and their downstream effects In particu-lar significant improvements and large-scale additions inthe areas of QSAR (quantitative structure activity rela-tionships) ADMET (absorption distribution metabol-ism excretion and toxicity) pharmacometabolomics andpharmacogenomics have been made Existing informationabout drug structures drug salt-forms drug names drugtargets and drug actions has also been expanded andupdated Numerous approved and experimental drugshave been added along with a number of new data fieldsdescribing each drug New search tools have also beendeveloped or improved on to increase the ease withwhich information can be foundMany of the enhancements made over the past 3 years
were stimulated by user feedback and suggestions We aregrateful to our users and continue to strive to meet theirneeds Further details on the additions and enhancementsmade to DrugBank 40 are described later
DATABASE ADDITIONS AND ENHANCEMENTS
The development and evolution of DrugBank includingprevious data content additions curation protocolsquality control methods general layout interfacefeatures and data sources has been described previously(8ndash10) Here we shall focus on enhancements and changesmade since the release of DrugBank 30 In particular wewill discuss (i) enhancements made to existing data (ii) theaddition of new data fields (iii) new and enhanced searchfeatures and (iv) a new user feedbackcommenting system
Enhancement of existing data
DrugBank has evolved significantly since its inception in2006 with a progressive expansion in the depth andbreadth of its data as well as improvements to thequality and reliability of its information The enhance-ments and additions in data content between DrugBank10 20 30 and 40 are summarized in Table 1 As seen inthis table DrugBank 40 has seen a further 40 increasein the number of data fields per DrugCard The number ofFDA-approved drugs has increased by 19 biotechdrugs by 14 and gt1200 new investigational drugs (inphase I II or III trials) have been added Likewise thenumber of drugs with transporters and carrier proteinshas increased by 40 and 30 respectively Additionallythe number of QSAR parameters has grown by 64 andthe number of ADMET parameters by 650 Thenumber of drugndashdrug interactions and the number ofdrugndashfood interactions have grown by 25 and 14respectively Similarly the number of drug target single-nucleotide polymorphisms (SNPs) has grown by 10whereas the number of SNPs associated with adversedrug effects (SNP-FX and SNP-ADR) has grown by78 In addition the number of illustrated drug actionpathways has grown by 38 Furthermore chemical syn-thesis information has been added to gt1200 drugs with
many synthesis references now hyperlinked to patents andreaction schemes maintained by Drugsynorg (httpwwwdrugsynorg)
Data quality and currency have always been two of thetop priorities for the DrugBank curation teamConsequently significant efforts over the past 3 yearshave been directed at improving the quality of existingDrugCards and keeping them as up-to-date as possibleUnlike DrugBank (ie the database) which has a 2ndash3-yearrelease cycle DrugBankrsquos DrugCards are continuouslyupdated as new information becomes available Becausenew drugs are constantly being approved and old drugsare perpetually being re-characterized re-purposed orremoved our curation team is continuously combing theliterature to update existing DrugCards or to add newDrugCards As a result hundreds of drug descriptionsdrug targets indications pharmacodynamics data mech-anisms of action and metabolism data fields have beenadded re-written andor expanded over the past 3 yearsWe have also updated and re-written all of DrugBankrsquosdrugndashdrug interactions In particular we have included orupdated all interactions where modifications in drugtherapy are recommended either by the manufacturer orother data sources
Drug nomenclature is particularly important forpharmacists physicians and the drug industry Manydrugs have dozens or even hundreds of alternate namesbrand names or synonyms often arising because ofvarious trademark patenting andor marketing require-ments For the latest release of DrugBank all of itssynonyms and brand names were extensively reviewedLengthy chemical and salt-form names were removedand only the most common synonyms have beenincluded enabling users to more easily browse alternatedrug names Brand names or synonyms appearing onmore than one DrugCard were also identified and editedor removed as required In total gt8000 duplicated or par-tially duplicated brand names and synonyms were editedor removed as part of DrugBank 40rsquos nomenclatureremediation effort
DrugBankrsquos experimental drug data set was alsoreviewed remediated and expanded In particular experi-mental drugs binding to nonndashhuman targets were removedwith the exception of those that bind to known bacterialviral and fungal pathogens Experimental drug targets ori-ginally referenced to the RCSB Protein Data Bank (5) arenow referenced to the original publications Additionalexperimental drugs were added from a variety of literaturesources and online databases such as ChEMBL (11)Further data quality checks on drug structures werecarried out by identifying InChI keys and ChEBI IDsappearing on more than one DrugCard In some casesduplicate IDs were the result of duplicate DrugCardswhereas in other cases the IDs referred to stereoisomersthat did not have stereochemistry depicted in their struc-tures These structures were manually corrected In total600 corrections of various forms were performed as partof the DrugBank 40 structure remediation effort
Drug classes are a useful way of grouping drugs withsimilar therapeutic effects and mechanisms of action Inthe latest version of DrugBank we expanded on the
2 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
information provided by the Anatomical TherapeuticChemical (ATC) Classification System defined by theWHO Collaborating Centre for Drug StatisticsMethodology (12) The first four levels of the classificationsystem (anatomical main group therapeutic subgrouppharmacological subgroup and chemical subgroup) arelisted for each drug with links to what we callCategoryCards The CategoryCard for each anatomicalmain group shows a listing of therapeutic subgroupswhich can be expanded to show pharmacological sub-groups which can be further expanded to show chemicalsubgroups which are expandable to finally show the fifthlevel chemical substances or drug names This new treestructure and the new browsing function called CategoryBrowse effectively replaces the old PharmaBrowse foundin earlier versions of DrugBank Adopting this well-estab-lished classification system in DrugBank 40 enables users tomore easily search for drugs with similar pharmacologiceffects
New data fields
As noted earlier significant efforts for this yearrsquos release ofDrugBank have been directed toward adding new data on
drug metabolism ADMET pharmacometabolomicspharmacogenomics and a variety of QSAR New datafields are also highlighted in Table 1 These additions toDrugBank 40 were motivated by the fast-moving devel-opments in the fields of xenobiotic metabolismpharmacometabolomics pharmacogenomics pre-clinicaldrug screening and computer-aided drug designWith regard to DrugBank 40rsquos effort to capture more
drug metabolism information it is important to note thatthe database now contains gt1200 drug metabolites(including their structures names activity abundanceand other detailed data) along with drug metabolismreaction data for gt1000 approved drugs These drug me-tabolism data include the sequence of reactions reactiontypes and whether drug metabolites are active inactiveandor major circulating entities In addition DrugBanknow offers comprehensive colorful manually drawninteractive drug metabolism pathways [using theSMPDB pathway model (13)] for gt50 drugs We believethat the addition of large numbers of drug metabolitesto DrugBank should also assist metabolomic andpharmacometabolomic studies by facilitating the identifi-cation of xenobiotics in biological samples (via MS
Table 1 Comparison between the coverage in DrugBank 10 20 30 and DrugBank 40
Category 10 20 30 40
Number of data fields per DrugCard 88 108 148 208Number of search types 8 12 16 18Number of illustrated drug-action pathways 0 0 168 232Number of drugs with metabolizing enzyme data 0 0 762 1037Number of drug metabolites with structuresa 0 0 0 1239Number of drug-metabolism reactionsa 0 0 0 1308Number of illustrated drug metabolism pathwaysa 0 0 0 53Number of drugs with drug transporter data 0 0 516 623Number of drugs with taxonomic classification informationa 0 0 0 6713Number of SNP-associated drug effects 0 0 113 201Number of drugs with patentpricingmanufacturer data 0 0 1208 1450Number of foodndashdrug interactions 0 714 1039 1180Number of drugndashdrug interactions 0 13 242 13 795 14 150Number of biochemical target-mediated drugndashdrug interactionsa 0 0 0 35 632Number of biochemical enzyme-mediated drugndashdrug interactionsa 0 0 0 226 463Number of biochemical transporter-mediated drugndashdrug interactionsa 0 0 0 60 644Number of biochemical drugndashdrug interactions (all)a 0 0 0 322 739Number of ADMET parameters (Caco-2 LogS) 0 276 890 6667Number of QSAR parameters per drug 5 6 14 23Number of drugs with drugndashtarget binding constant dataa 0 0 0 791Number of drugs with NMR spectraa 0 0 0 306Number of drugs with MS spectraa 0 0 0 384Number of drugs with chemical synthesis information 0 38 38 1285Number of FDA-approved small molecule drugs 841 1344 1424 1552Number of FDA small molecule drugs with salt-form structuresa 0 0 0 474Number of biotech drugs 113 123 132 284Number of nutraceutical drugs 61 69 82 87Number of withdrawn drugs 0 57 68 78Number of illicit drugs 0 188 189 190Number of experimental drugs 2894 3116 5210 6009Number of investigational drugs (phase I II and III trials)a 0 0 0 1219Total number of experimental and FDA small molecule drugs 3796 4774 6684 7561Total number of experimental and FDA drugs (all types) 3909 4897 6816 7713Number of all drug targets (unique) 2133 3037 4326 4115Number of approved-drug enzymescarriers (unique) 0 0 164 245Number of all drug enzymescarriers (unique) 0 0 169 253Number of external database links 12 18 31 33
aNew data
Nucleic Acids Research 2013 3
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
spectrometry) By providing the drug research communitywith gt1200 drug metabolites (along with their corres-ponding phase I and II enzymes) we are also hopingthat new and improved software could be developed forpredicting the structure of drug metabolites Efforts arealready underway in our laboratory to develop openaccess freeware for drug metabolism predictionIn addition to the substantial additions to DrugBankrsquos
drug metabolism data the quantity and diversity ofADMET data in DrugBank 40 has also been significantlyincreased Although a modest amount of ADMET datawas included in previous versions of DrugBank (8ndash10) itwas generally limited to a small number of approveddrugs However by taking advantage of several recentlyreleased software packages for predicting ADMETproperties (14) and mining newly published ADMETsurveys (1516) it has been possible to add 30 moreADMET data fields to nearly all of DrugBankrsquos drugentries These include measured andor predictedADMET values for Caco cell permeability (as a proxyfor intestinal permeability) bloodndashbrain barrier perme-ability human intestinal absorption levels P-glycoproteinactivity Cyp450 substrate preferences renal transportactivity carcinogenicity toxicity Ames test activity andhuman Ether a-go-go Related Gene (hERG) activityalong with their probability scoresAs the fields of metabolomics and pharmacome-
tabolomics continue to evolve we believe there is anincreasing need to develop drug databases that are com-patible with the needs of metabolomic researchers In par-ticular referential mass spectrometry (MS) and nuclearmagnetic resonance (NMR) spectra of pure drugs andtheir metabolites are particularly important for identifyingandor quantifying compounds in biological matricesPrescription and over-the-counter medications are fre-quently found in human biofluid samples characterizedin metabolomic studies To meet these compound identi-ficationcharacterization needs gt750 NMR and 3763 MSreference spectra for 400 different drugs have beenadded to DrugBank 40 Many of these spectra were ex-perimentally acquired by our laboratory (17) whereasothers have been assembled from freely available onlinesources (1819) In addition to including these referenceMS and NMR spectra we have also added a number ofadvanced spectral viewing and spectralmass matchingtools [originally developed for the HMDB (17)] to facili-tate metabolomic studies Using these tools users cansubmit chemical shift or mz (mass to charge ratio) listsand search against DrugBankrsquos spectral library for exactor even approximate matches Users can also browsezoom and view all of DrugBankrsquos spectral data tovisually compare results and assess spectral matches Tofurther enhance DrugBankrsquos compatibility withmetabolomic activities we have also added an extensivechemical taxonomy or chemical classification system thatuses the same hierarchical design and nomenclature asdeveloped for other metabolomic databases (1720) Inparticular compounds are classified into lsquoKingdomsrsquolsquoSuperclassesrsquo lsquoClassesrsquo lsquoSubclassesrsquo lsquoDirect ParentsrsquolsquoMolecular Frameworkrsquo lsquoSubstituentsrsquo and lsquoCross-Referencesrsquo using well-defined rules based on their
structural features and structure similarities This classifi-cation system allows users to identify all benzodiazepinesselect all alkaloids or extract all peptides in DrugBankusing a simple text query
To facilitate QSAR studies we have continued toadd to DrugBankrsquos collection of quantitative drug dataThis includes not only more molecular descriptors(number of H-bond donorsacceptors physiologicalcharge refractivity polarizability polar surface areaetc) but also more data on chemical substituents (seechemical taxonomy earlier) and extensive quantitative in-formation on drugndashligand binding constants The bindingconstant data have been mined from several sourcesincluding the primary literature and BindingDB (21)The combination of DrugBankrsquos existing 2D and 3Dstructural data along with new molecular descriptormolecular substituent and binding constant data for thou-sands of drugs should allow researchers to perform muchmore sophisticated and extensive QSAR studies
Many drugs exist in a variety of salt forms and thesesalt variants can play a significant role in the efficacydelivery or indication for certain drugs HistoricallyDrugBank has only provided the structure of lsquobasersquoform of the drug However to accommodate therequests of drug formulation experts and pharmacistswe have now added salt-form structures to DrugBank 40
New and enhanced search features
The quality of a database is determined not only by thequality of its content but also by the ease with which userscan access or navigate through its data In this regard theDrugBank team has continuously strived to improvethe databasersquos user interface to enhance its search utilitiesand to respond to user suggestions For DrugBank 40 wehave added two new browsing features to ease the retrievalof information (i) Reaction Browse and (ii) CategoryBrowse We have also improved the Multi-Searchfeature of DrugBankrsquos Interax (Drug InteractionLookup system) and streamlined the existing browsingutilities including DrugBrowse GenoBrowse andAssociation Browse A simple tool called BioInteractorfor rapidly retrieving and analyzing the biochemical datain DrugBank has also been added We have also upgradedDrugBankrsquos text search engine to improve the speed andquality of the search results and we have further enhancedDrugBankrsquos Data Extractor to make it easier to use andfar more accessible These new and improved searchbrowsing features should substantially increase the easewith which users are able to navigate retrieve andextract information from DrugBank
Reaction Browse was developed to facilitate the viewingand searching of drug metabolism reactions Its easilysearchable table format allows users to rapidly browsethrough pages displaying each of the drug metabolismreactions in DrugBank Each drug metabolite enzymeand reaction is hyperlinked to its respective DrugCardMetaboCard EnzymeCard and ReactionCard Searchbars above each column of the Reaction Browse tableenable users to search by drug name DrugBank ID me-tabolite name DrugBank metabolite ID enzyme name
4 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
UniProt ID or any combination of the above (Figure 1)Category Browse (which replaces the old PharmaBrowse)is a simple search tool that enables users to browse andsearch for drugs by category based on pharmacologicalaction or their ATC Classification (12) All four levels ofthe ATC classification system are listed together to facili-tate searching Each ATC drug category is hyperlinked toa CategoryCard as described earlier Furthermore eachdrug listed in the CategoryCard can be expanded toshow drug targets enabling users to easily compare drugtargets from the same category (Figure 1)
A new addition to DrugBankrsquos large selection ofsearch and browsing toolbox is called BioInteractorBioInteractor uses data on drug-target -enzyme and-transporter associations to provide insight on drugndashdrug interactions It allows users to identify forinstance which drugs act similarly on the same targetSuch drugs may have additive pharmacodynamic effectsif given concomitantly Conversely two drugs may antag-onize the beneficial effects of each other by havingopposite actions on a specific drug target Although clin-ically relevant drugndashdrug interactions are well docu-mented in commercial databases such as those producedby Lexi-Comp Inc the mechanisms of interactions arenot always clear BioInteractor uses ranked drugndashenzyme association information to predict enzyme-mediated drug interactions that may be clinicallyrelevant For this purpose enzyme inhibitors andinducers associated with DrugBankrsquos new drug metabol-ism data have been further categorized as lsquostrongrsquo lsquomod-eratersquo or lsquoneitherrsquo based on a set of criteria defined byBjornsson et al (22) Such a ranking system enablesusers to differentiate between drug effects on enzymeslikely to cause clinically significant changes to the pharma-codynamics of interacting drugs from those unlikelyto affect clinical outcomes A search query withBioInteractor will yield all possible enzyme-mediatedinteractions but the ranking system will provide furtherinsight on the probability of a clinically relevant inter-action occurring
In addition to these new search tools we have alsoimproved DrugBankrsquos standard search tools In particu-lar we have upgraded DrugBankrsquos text search engine touse Elasticsearch (httpwwwelasticsearchorg) anadvanced and scalable enterprise search backend It isalso now possible to search across drug targets drugndashdrug and drugndashfood interactions metabolic reactions me-tabolites and other relevant search fields In additionsearch results can now be downloaded using DrugBankrsquosupgraded Data Extractor The new Data Extractorappears in all of DrugBankrsquos standard browse andsearch windows allowing users to easily extract down-load and save their current data view at any point Inaddition all DrugBank data extractions performed byData Extractor are saved for up to 1 month so userscan return at any time to retrieve their results We havealso upgraded DrugBankrsquos Advanced Search thereby im-proving the granularity and coverage of data fields inDrugBank This should allow users to build much morecomplex queries through Advanced Searchrsquos simple inter-face DrugBankrsquos Advanced Search feature when
combined with the Data Extractor should allow users tosearch retrieve or extract almost any kind of data type orcombination of data within DrugBank
User feedback commenting system and news feed
Since the DrugBank was first released in 2006 the web hastransformed into a dynamic social environment Newsblogs and scientific online resources provide users withthe ability to leave comments and discuss articles Inresponse to these changes we have added a commentingfeature at the end of each DrugCard enabling users tosubmit comments that are moderated by the DrugBankcuration team This commenting system was designed toenable users to engage with the DrugBank curators toprovide a detailed record of changes and to stimulate con-versation among the scientific community Hundreds ofexcellent user comments have been received and actedon since this feature was implemented Additionally wehave created a DrugBank Twitter account allowingDrugBank curators to report news maintenance issuesand provide updates while at the same time allowingusers to supply feedback through social media Engagingour users has always been a priority and supplying avehicle to provide feedback will result in higher dataquality and an improved user experience
CONCLUSION
Over the past 7 years DrugBank has grown from a smallsomewhat specialized drug database to a large compre-hensive drug knowledgebase covering almost all aspects ofdrug function formulation and structure Throughout thisevolution we have strived to maintain the highest qualityand currency of data to continuously improve the qualityof its user interface and to attentively listen to all membersof its user community With this latest release ofDrugBank we have continued to make substantial im-provements to both the qualitycurrency of its data andthe quality of its user interface We have also added sub-stantial quantities of new and what we believe will behighly relevant data pertaining to drug metabolism drugmetabolites drug salt-forms ADMET and QSAR Theseadditions and enhancements are intended to facilitateresearch in xenobiotic metabolism (both prediction andcharacterization) pharmacometabolomics pharmacokin-etics pharmacodynamics pharmaceutics and drug designdiscovery They are also intended to make the data withinDrugBank relevant to a far wider audience including re-searchers in metabolomics pharmacogenomics andpharmacology formulation chemists pharmacists phys-icians educators and the general public Our commitmentto continue to maintain update and improve DrugBankwill continue for the foreseeable future Based on currenttrends in pharmaceutical and pharmacological research itis likely that future versions of DrugBank will includemany more drug action and drug metabolism pathwaysmuch more information regarding the transport anduptake of drugs as well as greater detail regarding themolecular mechanisms underlying drug toxicity anddrug-associated adverse reactions
Nucleic Acids Research 2013 5
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
information provided by the Anatomical TherapeuticChemical (ATC) Classification System defined by theWHO Collaborating Centre for Drug StatisticsMethodology (12) The first four levels of the classificationsystem (anatomical main group therapeutic subgrouppharmacological subgroup and chemical subgroup) arelisted for each drug with links to what we callCategoryCards The CategoryCard for each anatomicalmain group shows a listing of therapeutic subgroupswhich can be expanded to show pharmacological sub-groups which can be further expanded to show chemicalsubgroups which are expandable to finally show the fifthlevel chemical substances or drug names This new treestructure and the new browsing function called CategoryBrowse effectively replaces the old PharmaBrowse foundin earlier versions of DrugBank Adopting this well-estab-lished classification system in DrugBank 40 enables users tomore easily search for drugs with similar pharmacologiceffects
New data fields
As noted earlier significant efforts for this yearrsquos release ofDrugBank have been directed toward adding new data on
drug metabolism ADMET pharmacometabolomicspharmacogenomics and a variety of QSAR New datafields are also highlighted in Table 1 These additions toDrugBank 40 were motivated by the fast-moving devel-opments in the fields of xenobiotic metabolismpharmacometabolomics pharmacogenomics pre-clinicaldrug screening and computer-aided drug designWith regard to DrugBank 40rsquos effort to capture more
drug metabolism information it is important to note thatthe database now contains gt1200 drug metabolites(including their structures names activity abundanceand other detailed data) along with drug metabolismreaction data for gt1000 approved drugs These drug me-tabolism data include the sequence of reactions reactiontypes and whether drug metabolites are active inactiveandor major circulating entities In addition DrugBanknow offers comprehensive colorful manually drawninteractive drug metabolism pathways [using theSMPDB pathway model (13)] for gt50 drugs We believethat the addition of large numbers of drug metabolitesto DrugBank should also assist metabolomic andpharmacometabolomic studies by facilitating the identifi-cation of xenobiotics in biological samples (via MS
Table 1 Comparison between the coverage in DrugBank 10 20 30 and DrugBank 40
Category 10 20 30 40
Number of data fields per DrugCard 88 108 148 208Number of search types 8 12 16 18Number of illustrated drug-action pathways 0 0 168 232Number of drugs with metabolizing enzyme data 0 0 762 1037Number of drug metabolites with structuresa 0 0 0 1239Number of drug-metabolism reactionsa 0 0 0 1308Number of illustrated drug metabolism pathwaysa 0 0 0 53Number of drugs with drug transporter data 0 0 516 623Number of drugs with taxonomic classification informationa 0 0 0 6713Number of SNP-associated drug effects 0 0 113 201Number of drugs with patentpricingmanufacturer data 0 0 1208 1450Number of foodndashdrug interactions 0 714 1039 1180Number of drugndashdrug interactions 0 13 242 13 795 14 150Number of biochemical target-mediated drugndashdrug interactionsa 0 0 0 35 632Number of biochemical enzyme-mediated drugndashdrug interactionsa 0 0 0 226 463Number of biochemical transporter-mediated drugndashdrug interactionsa 0 0 0 60 644Number of biochemical drugndashdrug interactions (all)a 0 0 0 322 739Number of ADMET parameters (Caco-2 LogS) 0 276 890 6667Number of QSAR parameters per drug 5 6 14 23Number of drugs with drugndashtarget binding constant dataa 0 0 0 791Number of drugs with NMR spectraa 0 0 0 306Number of drugs with MS spectraa 0 0 0 384Number of drugs with chemical synthesis information 0 38 38 1285Number of FDA-approved small molecule drugs 841 1344 1424 1552Number of FDA small molecule drugs with salt-form structuresa 0 0 0 474Number of biotech drugs 113 123 132 284Number of nutraceutical drugs 61 69 82 87Number of withdrawn drugs 0 57 68 78Number of illicit drugs 0 188 189 190Number of experimental drugs 2894 3116 5210 6009Number of investigational drugs (phase I II and III trials)a 0 0 0 1219Total number of experimental and FDA small molecule drugs 3796 4774 6684 7561Total number of experimental and FDA drugs (all types) 3909 4897 6816 7713Number of all drug targets (unique) 2133 3037 4326 4115Number of approved-drug enzymescarriers (unique) 0 0 164 245Number of all drug enzymescarriers (unique) 0 0 169 253Number of external database links 12 18 31 33
aNew data
Nucleic Acids Research 2013 3
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
spectrometry) By providing the drug research communitywith gt1200 drug metabolites (along with their corres-ponding phase I and II enzymes) we are also hopingthat new and improved software could be developed forpredicting the structure of drug metabolites Efforts arealready underway in our laboratory to develop openaccess freeware for drug metabolism predictionIn addition to the substantial additions to DrugBankrsquos
drug metabolism data the quantity and diversity ofADMET data in DrugBank 40 has also been significantlyincreased Although a modest amount of ADMET datawas included in previous versions of DrugBank (8ndash10) itwas generally limited to a small number of approveddrugs However by taking advantage of several recentlyreleased software packages for predicting ADMETproperties (14) and mining newly published ADMETsurveys (1516) it has been possible to add 30 moreADMET data fields to nearly all of DrugBankrsquos drugentries These include measured andor predictedADMET values for Caco cell permeability (as a proxyfor intestinal permeability) bloodndashbrain barrier perme-ability human intestinal absorption levels P-glycoproteinactivity Cyp450 substrate preferences renal transportactivity carcinogenicity toxicity Ames test activity andhuman Ether a-go-go Related Gene (hERG) activityalong with their probability scoresAs the fields of metabolomics and pharmacome-
tabolomics continue to evolve we believe there is anincreasing need to develop drug databases that are com-patible with the needs of metabolomic researchers In par-ticular referential mass spectrometry (MS) and nuclearmagnetic resonance (NMR) spectra of pure drugs andtheir metabolites are particularly important for identifyingandor quantifying compounds in biological matricesPrescription and over-the-counter medications are fre-quently found in human biofluid samples characterizedin metabolomic studies To meet these compound identi-ficationcharacterization needs gt750 NMR and 3763 MSreference spectra for 400 different drugs have beenadded to DrugBank 40 Many of these spectra were ex-perimentally acquired by our laboratory (17) whereasothers have been assembled from freely available onlinesources (1819) In addition to including these referenceMS and NMR spectra we have also added a number ofadvanced spectral viewing and spectralmass matchingtools [originally developed for the HMDB (17)] to facili-tate metabolomic studies Using these tools users cansubmit chemical shift or mz (mass to charge ratio) listsand search against DrugBankrsquos spectral library for exactor even approximate matches Users can also browsezoom and view all of DrugBankrsquos spectral data tovisually compare results and assess spectral matches Tofurther enhance DrugBankrsquos compatibility withmetabolomic activities we have also added an extensivechemical taxonomy or chemical classification system thatuses the same hierarchical design and nomenclature asdeveloped for other metabolomic databases (1720) Inparticular compounds are classified into lsquoKingdomsrsquolsquoSuperclassesrsquo lsquoClassesrsquo lsquoSubclassesrsquo lsquoDirect ParentsrsquolsquoMolecular Frameworkrsquo lsquoSubstituentsrsquo and lsquoCross-Referencesrsquo using well-defined rules based on their
structural features and structure similarities This classifi-cation system allows users to identify all benzodiazepinesselect all alkaloids or extract all peptides in DrugBankusing a simple text query
To facilitate QSAR studies we have continued toadd to DrugBankrsquos collection of quantitative drug dataThis includes not only more molecular descriptors(number of H-bond donorsacceptors physiologicalcharge refractivity polarizability polar surface areaetc) but also more data on chemical substituents (seechemical taxonomy earlier) and extensive quantitative in-formation on drugndashligand binding constants The bindingconstant data have been mined from several sourcesincluding the primary literature and BindingDB (21)The combination of DrugBankrsquos existing 2D and 3Dstructural data along with new molecular descriptormolecular substituent and binding constant data for thou-sands of drugs should allow researchers to perform muchmore sophisticated and extensive QSAR studies
Many drugs exist in a variety of salt forms and thesesalt variants can play a significant role in the efficacydelivery or indication for certain drugs HistoricallyDrugBank has only provided the structure of lsquobasersquoform of the drug However to accommodate therequests of drug formulation experts and pharmacistswe have now added salt-form structures to DrugBank 40
New and enhanced search features
The quality of a database is determined not only by thequality of its content but also by the ease with which userscan access or navigate through its data In this regard theDrugBank team has continuously strived to improvethe databasersquos user interface to enhance its search utilitiesand to respond to user suggestions For DrugBank 40 wehave added two new browsing features to ease the retrievalof information (i) Reaction Browse and (ii) CategoryBrowse We have also improved the Multi-Searchfeature of DrugBankrsquos Interax (Drug InteractionLookup system) and streamlined the existing browsingutilities including DrugBrowse GenoBrowse andAssociation Browse A simple tool called BioInteractorfor rapidly retrieving and analyzing the biochemical datain DrugBank has also been added We have also upgradedDrugBankrsquos text search engine to improve the speed andquality of the search results and we have further enhancedDrugBankrsquos Data Extractor to make it easier to use andfar more accessible These new and improved searchbrowsing features should substantially increase the easewith which users are able to navigate retrieve andextract information from DrugBank
Reaction Browse was developed to facilitate the viewingand searching of drug metabolism reactions Its easilysearchable table format allows users to rapidly browsethrough pages displaying each of the drug metabolismreactions in DrugBank Each drug metabolite enzymeand reaction is hyperlinked to its respective DrugCardMetaboCard EnzymeCard and ReactionCard Searchbars above each column of the Reaction Browse tableenable users to search by drug name DrugBank ID me-tabolite name DrugBank metabolite ID enzyme name
4 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
UniProt ID or any combination of the above (Figure 1)Category Browse (which replaces the old PharmaBrowse)is a simple search tool that enables users to browse andsearch for drugs by category based on pharmacologicalaction or their ATC Classification (12) All four levels ofthe ATC classification system are listed together to facili-tate searching Each ATC drug category is hyperlinked toa CategoryCard as described earlier Furthermore eachdrug listed in the CategoryCard can be expanded toshow drug targets enabling users to easily compare drugtargets from the same category (Figure 1)
A new addition to DrugBankrsquos large selection ofsearch and browsing toolbox is called BioInteractorBioInteractor uses data on drug-target -enzyme and-transporter associations to provide insight on drugndashdrug interactions It allows users to identify forinstance which drugs act similarly on the same targetSuch drugs may have additive pharmacodynamic effectsif given concomitantly Conversely two drugs may antag-onize the beneficial effects of each other by havingopposite actions on a specific drug target Although clin-ically relevant drugndashdrug interactions are well docu-mented in commercial databases such as those producedby Lexi-Comp Inc the mechanisms of interactions arenot always clear BioInteractor uses ranked drugndashenzyme association information to predict enzyme-mediated drug interactions that may be clinicallyrelevant For this purpose enzyme inhibitors andinducers associated with DrugBankrsquos new drug metabol-ism data have been further categorized as lsquostrongrsquo lsquomod-eratersquo or lsquoneitherrsquo based on a set of criteria defined byBjornsson et al (22) Such a ranking system enablesusers to differentiate between drug effects on enzymeslikely to cause clinically significant changes to the pharma-codynamics of interacting drugs from those unlikelyto affect clinical outcomes A search query withBioInteractor will yield all possible enzyme-mediatedinteractions but the ranking system will provide furtherinsight on the probability of a clinically relevant inter-action occurring
In addition to these new search tools we have alsoimproved DrugBankrsquos standard search tools In particu-lar we have upgraded DrugBankrsquos text search engine touse Elasticsearch (httpwwwelasticsearchorg) anadvanced and scalable enterprise search backend It isalso now possible to search across drug targets drugndashdrug and drugndashfood interactions metabolic reactions me-tabolites and other relevant search fields In additionsearch results can now be downloaded using DrugBankrsquosupgraded Data Extractor The new Data Extractorappears in all of DrugBankrsquos standard browse andsearch windows allowing users to easily extract down-load and save their current data view at any point Inaddition all DrugBank data extractions performed byData Extractor are saved for up to 1 month so userscan return at any time to retrieve their results We havealso upgraded DrugBankrsquos Advanced Search thereby im-proving the granularity and coverage of data fields inDrugBank This should allow users to build much morecomplex queries through Advanced Searchrsquos simple inter-face DrugBankrsquos Advanced Search feature when
combined with the Data Extractor should allow users tosearch retrieve or extract almost any kind of data type orcombination of data within DrugBank
User feedback commenting system and news feed
Since the DrugBank was first released in 2006 the web hastransformed into a dynamic social environment Newsblogs and scientific online resources provide users withthe ability to leave comments and discuss articles Inresponse to these changes we have added a commentingfeature at the end of each DrugCard enabling users tosubmit comments that are moderated by the DrugBankcuration team This commenting system was designed toenable users to engage with the DrugBank curators toprovide a detailed record of changes and to stimulate con-versation among the scientific community Hundreds ofexcellent user comments have been received and actedon since this feature was implemented Additionally wehave created a DrugBank Twitter account allowingDrugBank curators to report news maintenance issuesand provide updates while at the same time allowingusers to supply feedback through social media Engagingour users has always been a priority and supplying avehicle to provide feedback will result in higher dataquality and an improved user experience
CONCLUSION
Over the past 7 years DrugBank has grown from a smallsomewhat specialized drug database to a large compre-hensive drug knowledgebase covering almost all aspects ofdrug function formulation and structure Throughout thisevolution we have strived to maintain the highest qualityand currency of data to continuously improve the qualityof its user interface and to attentively listen to all membersof its user community With this latest release ofDrugBank we have continued to make substantial im-provements to both the qualitycurrency of its data andthe quality of its user interface We have also added sub-stantial quantities of new and what we believe will behighly relevant data pertaining to drug metabolism drugmetabolites drug salt-forms ADMET and QSAR Theseadditions and enhancements are intended to facilitateresearch in xenobiotic metabolism (both prediction andcharacterization) pharmacometabolomics pharmacokin-etics pharmacodynamics pharmaceutics and drug designdiscovery They are also intended to make the data withinDrugBank relevant to a far wider audience including re-searchers in metabolomics pharmacogenomics andpharmacology formulation chemists pharmacists phys-icians educators and the general public Our commitmentto continue to maintain update and improve DrugBankwill continue for the foreseeable future Based on currenttrends in pharmaceutical and pharmacological research itis likely that future versions of DrugBank will includemany more drug action and drug metabolism pathwaysmuch more information regarding the transport anduptake of drugs as well as greater detail regarding themolecular mechanisms underlying drug toxicity anddrug-associated adverse reactions
Nucleic Acids Research 2013 5
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
spectrometry) By providing the drug research communitywith gt1200 drug metabolites (along with their corres-ponding phase I and II enzymes) we are also hopingthat new and improved software could be developed forpredicting the structure of drug metabolites Efforts arealready underway in our laboratory to develop openaccess freeware for drug metabolism predictionIn addition to the substantial additions to DrugBankrsquos
drug metabolism data the quantity and diversity ofADMET data in DrugBank 40 has also been significantlyincreased Although a modest amount of ADMET datawas included in previous versions of DrugBank (8ndash10) itwas generally limited to a small number of approveddrugs However by taking advantage of several recentlyreleased software packages for predicting ADMETproperties (14) and mining newly published ADMETsurveys (1516) it has been possible to add 30 moreADMET data fields to nearly all of DrugBankrsquos drugentries These include measured andor predictedADMET values for Caco cell permeability (as a proxyfor intestinal permeability) bloodndashbrain barrier perme-ability human intestinal absorption levels P-glycoproteinactivity Cyp450 substrate preferences renal transportactivity carcinogenicity toxicity Ames test activity andhuman Ether a-go-go Related Gene (hERG) activityalong with their probability scoresAs the fields of metabolomics and pharmacome-
tabolomics continue to evolve we believe there is anincreasing need to develop drug databases that are com-patible with the needs of metabolomic researchers In par-ticular referential mass spectrometry (MS) and nuclearmagnetic resonance (NMR) spectra of pure drugs andtheir metabolites are particularly important for identifyingandor quantifying compounds in biological matricesPrescription and over-the-counter medications are fre-quently found in human biofluid samples characterizedin metabolomic studies To meet these compound identi-ficationcharacterization needs gt750 NMR and 3763 MSreference spectra for 400 different drugs have beenadded to DrugBank 40 Many of these spectra were ex-perimentally acquired by our laboratory (17) whereasothers have been assembled from freely available onlinesources (1819) In addition to including these referenceMS and NMR spectra we have also added a number ofadvanced spectral viewing and spectralmass matchingtools [originally developed for the HMDB (17)] to facili-tate metabolomic studies Using these tools users cansubmit chemical shift or mz (mass to charge ratio) listsand search against DrugBankrsquos spectral library for exactor even approximate matches Users can also browsezoom and view all of DrugBankrsquos spectral data tovisually compare results and assess spectral matches Tofurther enhance DrugBankrsquos compatibility withmetabolomic activities we have also added an extensivechemical taxonomy or chemical classification system thatuses the same hierarchical design and nomenclature asdeveloped for other metabolomic databases (1720) Inparticular compounds are classified into lsquoKingdomsrsquolsquoSuperclassesrsquo lsquoClassesrsquo lsquoSubclassesrsquo lsquoDirect ParentsrsquolsquoMolecular Frameworkrsquo lsquoSubstituentsrsquo and lsquoCross-Referencesrsquo using well-defined rules based on their
structural features and structure similarities This classifi-cation system allows users to identify all benzodiazepinesselect all alkaloids or extract all peptides in DrugBankusing a simple text query
To facilitate QSAR studies we have continued toadd to DrugBankrsquos collection of quantitative drug dataThis includes not only more molecular descriptors(number of H-bond donorsacceptors physiologicalcharge refractivity polarizability polar surface areaetc) but also more data on chemical substituents (seechemical taxonomy earlier) and extensive quantitative in-formation on drugndashligand binding constants The bindingconstant data have been mined from several sourcesincluding the primary literature and BindingDB (21)The combination of DrugBankrsquos existing 2D and 3Dstructural data along with new molecular descriptormolecular substituent and binding constant data for thou-sands of drugs should allow researchers to perform muchmore sophisticated and extensive QSAR studies
Many drugs exist in a variety of salt forms and thesesalt variants can play a significant role in the efficacydelivery or indication for certain drugs HistoricallyDrugBank has only provided the structure of lsquobasersquoform of the drug However to accommodate therequests of drug formulation experts and pharmacistswe have now added salt-form structures to DrugBank 40
New and enhanced search features
The quality of a database is determined not only by thequality of its content but also by the ease with which userscan access or navigate through its data In this regard theDrugBank team has continuously strived to improvethe databasersquos user interface to enhance its search utilitiesand to respond to user suggestions For DrugBank 40 wehave added two new browsing features to ease the retrievalof information (i) Reaction Browse and (ii) CategoryBrowse We have also improved the Multi-Searchfeature of DrugBankrsquos Interax (Drug InteractionLookup system) and streamlined the existing browsingutilities including DrugBrowse GenoBrowse andAssociation Browse A simple tool called BioInteractorfor rapidly retrieving and analyzing the biochemical datain DrugBank has also been added We have also upgradedDrugBankrsquos text search engine to improve the speed andquality of the search results and we have further enhancedDrugBankrsquos Data Extractor to make it easier to use andfar more accessible These new and improved searchbrowsing features should substantially increase the easewith which users are able to navigate retrieve andextract information from DrugBank
Reaction Browse was developed to facilitate the viewingand searching of drug metabolism reactions Its easilysearchable table format allows users to rapidly browsethrough pages displaying each of the drug metabolismreactions in DrugBank Each drug metabolite enzymeand reaction is hyperlinked to its respective DrugCardMetaboCard EnzymeCard and ReactionCard Searchbars above each column of the Reaction Browse tableenable users to search by drug name DrugBank ID me-tabolite name DrugBank metabolite ID enzyme name
4 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
UniProt ID or any combination of the above (Figure 1)Category Browse (which replaces the old PharmaBrowse)is a simple search tool that enables users to browse andsearch for drugs by category based on pharmacologicalaction or their ATC Classification (12) All four levels ofthe ATC classification system are listed together to facili-tate searching Each ATC drug category is hyperlinked toa CategoryCard as described earlier Furthermore eachdrug listed in the CategoryCard can be expanded toshow drug targets enabling users to easily compare drugtargets from the same category (Figure 1)
A new addition to DrugBankrsquos large selection ofsearch and browsing toolbox is called BioInteractorBioInteractor uses data on drug-target -enzyme and-transporter associations to provide insight on drugndashdrug interactions It allows users to identify forinstance which drugs act similarly on the same targetSuch drugs may have additive pharmacodynamic effectsif given concomitantly Conversely two drugs may antag-onize the beneficial effects of each other by havingopposite actions on a specific drug target Although clin-ically relevant drugndashdrug interactions are well docu-mented in commercial databases such as those producedby Lexi-Comp Inc the mechanisms of interactions arenot always clear BioInteractor uses ranked drugndashenzyme association information to predict enzyme-mediated drug interactions that may be clinicallyrelevant For this purpose enzyme inhibitors andinducers associated with DrugBankrsquos new drug metabol-ism data have been further categorized as lsquostrongrsquo lsquomod-eratersquo or lsquoneitherrsquo based on a set of criteria defined byBjornsson et al (22) Such a ranking system enablesusers to differentiate between drug effects on enzymeslikely to cause clinically significant changes to the pharma-codynamics of interacting drugs from those unlikelyto affect clinical outcomes A search query withBioInteractor will yield all possible enzyme-mediatedinteractions but the ranking system will provide furtherinsight on the probability of a clinically relevant inter-action occurring
In addition to these new search tools we have alsoimproved DrugBankrsquos standard search tools In particu-lar we have upgraded DrugBankrsquos text search engine touse Elasticsearch (httpwwwelasticsearchorg) anadvanced and scalable enterprise search backend It isalso now possible to search across drug targets drugndashdrug and drugndashfood interactions metabolic reactions me-tabolites and other relevant search fields In additionsearch results can now be downloaded using DrugBankrsquosupgraded Data Extractor The new Data Extractorappears in all of DrugBankrsquos standard browse andsearch windows allowing users to easily extract down-load and save their current data view at any point Inaddition all DrugBank data extractions performed byData Extractor are saved for up to 1 month so userscan return at any time to retrieve their results We havealso upgraded DrugBankrsquos Advanced Search thereby im-proving the granularity and coverage of data fields inDrugBank This should allow users to build much morecomplex queries through Advanced Searchrsquos simple inter-face DrugBankrsquos Advanced Search feature when
combined with the Data Extractor should allow users tosearch retrieve or extract almost any kind of data type orcombination of data within DrugBank
User feedback commenting system and news feed
Since the DrugBank was first released in 2006 the web hastransformed into a dynamic social environment Newsblogs and scientific online resources provide users withthe ability to leave comments and discuss articles Inresponse to these changes we have added a commentingfeature at the end of each DrugCard enabling users tosubmit comments that are moderated by the DrugBankcuration team This commenting system was designed toenable users to engage with the DrugBank curators toprovide a detailed record of changes and to stimulate con-versation among the scientific community Hundreds ofexcellent user comments have been received and actedon since this feature was implemented Additionally wehave created a DrugBank Twitter account allowingDrugBank curators to report news maintenance issuesand provide updates while at the same time allowingusers to supply feedback through social media Engagingour users has always been a priority and supplying avehicle to provide feedback will result in higher dataquality and an improved user experience
CONCLUSION
Over the past 7 years DrugBank has grown from a smallsomewhat specialized drug database to a large compre-hensive drug knowledgebase covering almost all aspects ofdrug function formulation and structure Throughout thisevolution we have strived to maintain the highest qualityand currency of data to continuously improve the qualityof its user interface and to attentively listen to all membersof its user community With this latest release ofDrugBank we have continued to make substantial im-provements to both the qualitycurrency of its data andthe quality of its user interface We have also added sub-stantial quantities of new and what we believe will behighly relevant data pertaining to drug metabolism drugmetabolites drug salt-forms ADMET and QSAR Theseadditions and enhancements are intended to facilitateresearch in xenobiotic metabolism (both prediction andcharacterization) pharmacometabolomics pharmacokin-etics pharmacodynamics pharmaceutics and drug designdiscovery They are also intended to make the data withinDrugBank relevant to a far wider audience including re-searchers in metabolomics pharmacogenomics andpharmacology formulation chemists pharmacists phys-icians educators and the general public Our commitmentto continue to maintain update and improve DrugBankwill continue for the foreseeable future Based on currenttrends in pharmaceutical and pharmacological research itis likely that future versions of DrugBank will includemany more drug action and drug metabolism pathwaysmuch more information regarding the transport anduptake of drugs as well as greater detail regarding themolecular mechanisms underlying drug toxicity anddrug-associated adverse reactions
Nucleic Acids Research 2013 5
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
UniProt ID or any combination of the above (Figure 1)Category Browse (which replaces the old PharmaBrowse)is a simple search tool that enables users to browse andsearch for drugs by category based on pharmacologicalaction or their ATC Classification (12) All four levels ofthe ATC classification system are listed together to facili-tate searching Each ATC drug category is hyperlinked toa CategoryCard as described earlier Furthermore eachdrug listed in the CategoryCard can be expanded toshow drug targets enabling users to easily compare drugtargets from the same category (Figure 1)
A new addition to DrugBankrsquos large selection ofsearch and browsing toolbox is called BioInteractorBioInteractor uses data on drug-target -enzyme and-transporter associations to provide insight on drugndashdrug interactions It allows users to identify forinstance which drugs act similarly on the same targetSuch drugs may have additive pharmacodynamic effectsif given concomitantly Conversely two drugs may antag-onize the beneficial effects of each other by havingopposite actions on a specific drug target Although clin-ically relevant drugndashdrug interactions are well docu-mented in commercial databases such as those producedby Lexi-Comp Inc the mechanisms of interactions arenot always clear BioInteractor uses ranked drugndashenzyme association information to predict enzyme-mediated drug interactions that may be clinicallyrelevant For this purpose enzyme inhibitors andinducers associated with DrugBankrsquos new drug metabol-ism data have been further categorized as lsquostrongrsquo lsquomod-eratersquo or lsquoneitherrsquo based on a set of criteria defined byBjornsson et al (22) Such a ranking system enablesusers to differentiate between drug effects on enzymeslikely to cause clinically significant changes to the pharma-codynamics of interacting drugs from those unlikelyto affect clinical outcomes A search query withBioInteractor will yield all possible enzyme-mediatedinteractions but the ranking system will provide furtherinsight on the probability of a clinically relevant inter-action occurring
In addition to these new search tools we have alsoimproved DrugBankrsquos standard search tools In particu-lar we have upgraded DrugBankrsquos text search engine touse Elasticsearch (httpwwwelasticsearchorg) anadvanced and scalable enterprise search backend It isalso now possible to search across drug targets drugndashdrug and drugndashfood interactions metabolic reactions me-tabolites and other relevant search fields In additionsearch results can now be downloaded using DrugBankrsquosupgraded Data Extractor The new Data Extractorappears in all of DrugBankrsquos standard browse andsearch windows allowing users to easily extract down-load and save their current data view at any point Inaddition all DrugBank data extractions performed byData Extractor are saved for up to 1 month so userscan return at any time to retrieve their results We havealso upgraded DrugBankrsquos Advanced Search thereby im-proving the granularity and coverage of data fields inDrugBank This should allow users to build much morecomplex queries through Advanced Searchrsquos simple inter-face DrugBankrsquos Advanced Search feature when
combined with the Data Extractor should allow users tosearch retrieve or extract almost any kind of data type orcombination of data within DrugBank
User feedback commenting system and news feed
Since the DrugBank was first released in 2006 the web hastransformed into a dynamic social environment Newsblogs and scientific online resources provide users withthe ability to leave comments and discuss articles Inresponse to these changes we have added a commentingfeature at the end of each DrugCard enabling users tosubmit comments that are moderated by the DrugBankcuration team This commenting system was designed toenable users to engage with the DrugBank curators toprovide a detailed record of changes and to stimulate con-versation among the scientific community Hundreds ofexcellent user comments have been received and actedon since this feature was implemented Additionally wehave created a DrugBank Twitter account allowingDrugBank curators to report news maintenance issuesand provide updates while at the same time allowingusers to supply feedback through social media Engagingour users has always been a priority and supplying avehicle to provide feedback will result in higher dataquality and an improved user experience
CONCLUSION
Over the past 7 years DrugBank has grown from a smallsomewhat specialized drug database to a large compre-hensive drug knowledgebase covering almost all aspects ofdrug function formulation and structure Throughout thisevolution we have strived to maintain the highest qualityand currency of data to continuously improve the qualityof its user interface and to attentively listen to all membersof its user community With this latest release ofDrugBank we have continued to make substantial im-provements to both the qualitycurrency of its data andthe quality of its user interface We have also added sub-stantial quantities of new and what we believe will behighly relevant data pertaining to drug metabolism drugmetabolites drug salt-forms ADMET and QSAR Theseadditions and enhancements are intended to facilitateresearch in xenobiotic metabolism (both prediction andcharacterization) pharmacometabolomics pharmacokin-etics pharmacodynamics pharmaceutics and drug designdiscovery They are also intended to make the data withinDrugBank relevant to a far wider audience including re-searchers in metabolomics pharmacogenomics andpharmacology formulation chemists pharmacists phys-icians educators and the general public Our commitmentto continue to maintain update and improve DrugBankwill continue for the foreseeable future Based on currenttrends in pharmaceutical and pharmacological research itis likely that future versions of DrugBank will includemany more drug action and drug metabolism pathwaysmuch more information regarding the transport anduptake of drugs as well as greater detail regarding themolecular mechanisms underlying drug toxicity anddrug-associated adverse reactions
Nucleic Acids Research 2013 5
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
Figure 1 A screenshot montage of some of DrugBank 40rsquos new browsing tools and spectral search features
6 Nucleic Acids Research 2013
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from
ACKNOWLEDGEMENTS
The authors would like to thank Allison Pon and KellyBanco for their contributions and the many users ofDrugBank for their valuable feedback and suggestions
FUNDING
The authors wish to thank Genome Alberta (a division ofGenome Canada) The Canadian Institutes of HealthResearch (CIHR) and Alberta Innovates HealthSolutions (AIHS) for financial support Funding foropen access charge CIHR
Conflict of interest statement None declared
REFERENCES
1 ThornCF KleinTE and AltmanRB (2013) PharmGKB thepharmacogenomics knowledge base Methods Mol Biol 1015311ndash320
2 HastingsJ de MatosP DekkerA EnnisM HarshaBKaleN MuthukrishnanV OwenG TurnerS WilliamsMet al (2013) The ChEBI reference database and ontology forbiologically relevant chemistry enhancements for 2013 NucleicAcids Res 41 D456ndashD463
3 KanehisaM GotoS SatoY FurumichiM and TanabeM(2012) KEGG for integration and interpretation of large-scalemolecular data sets Nucleic Acids Res 40 D109ndashD114
4 StelzerG DalahI SteinTI SatanowerY RosenN NativNOz-LeviD OlenderT BelinkyF BahirI et al (2011)In-silico human genomics with GeneCards Hum Genomics 5709ndash717
5 RosePW BiC BluhmWF ChristieCH DimitropoulosDDuttaS GreenRK GoodsellDS PrlicA QuesadaM et al(2013) The RCSB protein data bank new resources for researchand education Nucleic Acids Res 41 D475ndashD482
6 NCBI Resource Coordinators (2013) Database resources of thenational center for biotechnology information Nucleic Acids Res41 D8ndashD20
7 UniProt Consortium (2013) Update on activities at the UniversalProtein Resource (UniProt) in 2013 Nucleic Acids Res 41D43ndashD47
8 WishartDS KnoxC GuoAC ShirvastavaS HassanaliMStothardP ChangZ and WoolseyJ (2006) DrugBank acomprehensive resources for in silico drug discovery andexploration Nucleic Acids Res 34 D668ndashD672
9 WishartDS KnoxC GuoAC ChengD ShrivastavaSTzurD GautamB and HassanaliM (2008) DrugBank aknowledgebase for drugs drug actions and drug targets NucleicAcids Res 36 D901ndashD906
10 KnoxC LawV JewisonT LiuP LyS FrolkisA PonABancoK MakC NeveuV et al (2011) DrugBank 30 acomprehensive resources for lsquoOmicsrsquo research on drugs NucleicAcids Res 39 D1035ndashD1041
11 GaultonA BellisLJ BentoAP ChambersJ DaviesMHerseyA LightY McGlincheyS MichalovichDAl-LazikaniB et al (2011) ChEMBL a large-scale bioactivitydatabase for drug discovery Nucleic Acids Res 40D1100ndashD1107
12 WHO Collaborating Centre for Drug Statistics MethodologyATC classification index with DDDs 2013 Oslo 2012
13 FrolkisA KnoxC LimE JewisonT LawV HauDDLiuP GautamB LyS GuoAC et al (2010) SMPDB thesmall molecule pathway database Nucleic Acids Res 38D480ndashD487
14 ChengF LiW ZhouY ShenJ WuZ LiuG LeePW andTangY (2012) admetSAR a comprehensive source and free toolfor assessment of chemical ADMET properties J Chem InfModel 52 3099ndash3105
15 CaoD WangJ ZhouR LiY YuH and HouT (2012)ADMET evaluation in drug discovery 11 PharmacoKineticsKnowledge Base (PKKB) a comprehensive database ofpharmacokinetic and toxic properties for drugs J Chem InfModel 52 1132ndash1137
16 ModaTL TorresLG CarraraAE and AndricopuloAD(2008) PKDB database for pharmacokinetic propertiesand predictive in silico ADME models Bioinformatics 242270ndash2271
17 WishartDS JewisonT GuoAC WilsonM KnoxC LiuYDjoumbouY MandalR AziatF DongE et al (2013) HMDB30-the human metabolome database in 2013 Nucleic Acids Res41 D801ndashD807
18 UlrichEL AkutsuH DoreleijersJF HaranoYIoannidisYE LinJ LivnyM MadingS MaziukD MillerZet al (2008) BioMagResBank Nucleic Acids Res 36D402ndashD408
19 KopkaJ SchauerN KruegerS BirkemeyerC UsadelBBergmullerE DormannP WeckwerthW GibonY StittMet al (2005) GMDCSBDB the golm metabolome databaseBioinformatics 21 1635ndash1638
20 JewisonT WilsonM LiuY KnoxC DjoumbouY LoPMandalR KrishnamurthyR and WishartDS (2013) ECMDBthe E coli metabolome database Nucleic Acids Res 41D625ndashD630
21 LiuT LinY WenX JorissenRN and GilsonMK (2007)BindingDB a web-accessible database of experimentallydetermined protein-ligand binding affinities Nucleic Acids Res35 D198ndashD201
22 BjornssonTD CallaghanJT EinolfHJ FischerV GanLGrimmS KaoJ KingSP MiwaG NiL et al (2003)The conduct of in vitro and in vivo drug-drug interactionstudies a PhRMA perspective J Clin Pharmacol 43443ndash469
Nucleic Acids Research 2013 7
at University of A
lberta on Novem
ber 25 2013httpnaroxfordjournalsorg
Dow
nloaded from