Should medicinal chemists do molecular modelling?

1 DRUDIS 961 1 4 Please cite this article in press as: Ritchie, Should medicinal chemists do molecular modelling ?, Drug Discov Today (2012), medicinal chemists domolecular modelling ?Timothy J. RitchieQ11, and Iain M. McLay2In this article we discuss the pros and cons of medicinal chemists undertaking three-dimensional (3D)computer-aidedQ2drug design (CADD) activities for themselves, from the viewpoint of both medicinalchemists and computational chemists . We describe how best this can be implemented, the potentialbenefits that can be obtained and the pitfalls that are often is not common practice for a computationalchemist or molecular modeller to leave his or herworkstation, don a white coat and performorganic chemistry experiments in a medicinalchemistry lab. Indeed many people would con-sider this to be rather reckless behaviour. Con-versely there has always been a school ofthought that medicinal chemists Should carryout, to a greater or lesser extent, some elementsof computer-aided drug design (CADD) forthemselves, but is this suggestion just as reck-less?

2 In this article we discuss the reasoningbehind this approach, how it could be imple-mented and the potential advantages and pit-falls that can accompany this chemists use computer programs ona daily basis during the course of their work, forexample to generate experiments in an electroniclaboratory notebook, perform database searches,retrieve biological test results, calculate physico-chemical properties and generate plots toexamine structure activity relationships. Whenchemical structures are involved, they are usuallyconsidered in two the purposes of this article, the definitionof CADD will not cover such activities, but ratherfocus on computational approaches usingthree-dimensional (3D) molecular structures,such as conformational analysis, molecularoverlays, pharmacophore modelling andsearching, ligand protein docking, amongothers. A comprehensive review of CADD-related software and resources has recentlybeen published [1] in addition to a discussionaround the general topic of specialist andmainstream informatics and computationaltools for drug discovery [2].

3 Do medicinal chemists want to do CADD?The extent to which medicinal chemists want todo CADD for themselves varies three groups can be identified:(i) Enthusiastic: this group proactively seeksways to get involved in CADD-relatedactivities, often becoming local experts in their laboratories or departments.(ii) Curious: this group is not performing suchtasks but appear to be interested in doing so;often they are not clear about what optionsare available or who to approach to initiatethe interaction. This group is likely to needhelp and encouragement from CADD expertsto become comfortable with using CADD approaches.(iii) Sceptical: this group tends to be uninter-ested in or even dismissive of CADD andprefer to rely on traditional medicinalchemistry approaches. It is debatablewhether this group Should be coerced into performing CADD, but a powerfulstimulus perhaps would be observing theircontemporaries using CADD approachessuccessfully in their discovery computational chemists wantmedicinal chemists to do CADD?

4 Typically, computational chemists fall into fourdistinct groupings when this topic is discussed:(i) Enthusiastic: they believe CADD techniquescan be a significant aid to the creativity ofthe medicinal chemist and that everythingpossible Should be conducted to get robusttechniques into their hands.(ii) Luke warm: they have the opinion that if amedicinal chemist wants to play withCADD techniques they can be assisted,but it would be best not to encouragethem.(iii) Hostile: they consider modelling performedby medicinal chemists to be generallyworthless and also dangerous as it canFeatures PERSPECTIVEDrug Discovery Today Volume 00, Number 00 January 2012 PERSPECTIVE1359-6446/06/$ - see front matter 2012 Published by Elsevier Ltd. 1generate time-consuming synthetic workbased on misunderstandings of how CADD techniques Should be used and how resultsshould be interpreted.(iv) Scared: if medicinal chemists know how toperform CADD techniques they will nolonger need professional software vendors want medicinalchemists to do CADD?

5 It is perhaps obvious that companies sellingCADD-related software applications would likemedicinal chemists to use their products toincrease sales and their user base. In recent yearsmany molecular modelling software companieshave moved significantly into areas, such aschemical information and data management(often through mergers), but they retain a stronginterest in expanding usage among the medic-inal chemists . To this end, many companiesincluding Accelrys ( ), Chemi-cal Computing Group ( ), Schrodinger ( ), OpenEye ( ) and Tripos ( ) actively promote applicationsthat are specifically designed to target themedicinal chemistry end-user, often using terms,such as streamlined , simplified , and user-friendly , to entice medicinal chemists who mightbe put off by full-featured CADD companies have even been known toattempt to bypass the CADD department com-pletely, appealing directly to the medicinalchemists as potential customers. If the interfacesto these tools are well designed (ideally withinput from real medicinal chemists ) thisapproach can be useful and indeed has beenimplemented successfully in many companies.

6 Itis important, however, that the underlyingcomputational methodologies Should not alsobe streamlined to the point where the com-putational chemist views them as inferior tothose available in the fully featured the best approach is to enable medic-inal chemists to use a simplified interface toapplications that are also used by addition to software suppliers, there areother groups that produce CADD-related appli-cations aimed at the medicinal chemist , sug-gesting that there is a perceived need for suchtools: molecular ConceptorTMis an set of e-learning modules that aim to educate themedicinal chemist in aspects of drug design [3]; arecent publication by an academic groupdescribes a PyMOL-based CADD platform formedicinal chemists [4]. Furthermore, reviewarticles on CADD-related topics that targetmedicinal chemists have been published in theliterature, such as one covering molecularinteractions [5].Approaches used to encourage medicinalchemists to undertake CADD activitiesIf a decision is made to encourage medicinalchemists to undertake CADD for themselves,there are various strategies that can be used totry to achieve this goal (Fig.)

7 1):(i) Put workstations into every chemistrylaboratory, and send all chemists on atraining course for modelling software;many pharmaceutical companies have triedthis. The approach could fail if it is notthought through carefully enough: who willmonitor the work that is subsequentlycarried out? And will CADD hypotheses thatare formulated be actually tested bysynthesising the suggested target mole-cules? If the medicinal chemists do not usethe CADD software regularly enough or ifthere is insufficient on-going support fromCADD experts, the workstation might notbe maintained and will eventually fall intodisuse, leaving an expensive piece ofequipment gathering dust at an emptydesk. The approach is more successful if amedicinal chemist is instructed to dedicatea proportion of the working day to suchwork with end of year targets set formodelling activities. If this is not done themedicinal chemist inevitably will be suckedtowards the fume hood to prepare com-pounds that will satisfy end of yearsynthesis targets.

8 (ii) Create simple, web-based interfaces toCADD tools, to increase user-friendlinessand also control what functionality themedicinal chemist is exposed to. This is anattractive approach and many companiesincluding Novartis ( ) [6], GlaxoSmithKline( ) [7], Abbott ( ) [8] and Pfizer( )[ modelling on the web: enabling chemiststo pursue drug design calculations: ]have implemented such tools with con-siderable success: medicinal chemists areable to get meaningful results quickly in afew simple steps without needing to knowin detail what is going on in the back-ground. It is important that such applica-tions have a well-thought-out interface,adequate documentation and are neithertoo simplistic nor too complicated.(iii) Extract information from more complexCADD experiments and generate staticdatabases that can be accessed by medic-inal chemists . This is related to the pointabove and enables chemists to accesscomputational results that have beenpreviously validated by experts.

9 Typicallythese would be tools to provide access toprotein ligand crystal structures, dockingresults or molecular alignment results. ThePERSPECTIVE Drug Discovery Today Volume 00, Number 00 January 2012 DRUDIS 961 1 4 Please cite this article in press as: Ritchie, Should medicinal chemists do molecular modelling ?, Drug Discov Today (2012), workstations in all laborato-ries and provide CADD training tomedicinal chemistsProvide useful CADD applicationsas simplified web toolsPrepackage CADD results for easyviewing and analysisEmbed modeller within eachmedicinal chemistry teamEmploy expert as a bridgebetween computational andmedicinal chemistry groupsHigh start-up costs andcontinuous effort required tomaintain competenceRequires development of intuitive,robust user interfacesNeeds agreement on format andlevel of detail to be suppliedModeller disconnected from maincomputational chemistry groupManagement agreement requiredfor dedicated headcount - difficultin smaller companiesDeploy workstation s in a ll laborato-ries and pro vide CADD training t omedicinal chemi stsProvide useful CADD appl icationsas simpli fied web toolsPrepackage C ADD resul ts for easyviewing and ana lysisEmbed modeller within eachmedicinal chemi stry teamEmploy exper t as a bridg ebetween computational an dmedicinal chemi stry groupsStrategiesHigh start-u p costs andcontin uous effort requir ed tomaintain competenceRequires developme nt of intuitive.

10 Robust user interf acesNeeds agreement on forma t andlevel of detail to be suppliedModeller disconnected from maincomputational c hemistry groupManagemen t agreeme nt requiredfor dedic ated headcount - difficultin smaller companiesChallengesDrug Discovery Today FIGURE 1 Approaches to encourage medicinal chemists to undertake computer-aided drug design (CADD):strategies and PERSPECTIVE default views would be agreed by eachproject team and the medicinal chemistwould be presented with tools to enablesimple exploration of the ligand proteininteractions and the measurement ofgeometric properties, among others.(iv) Embed a computational chemist (full- orpart-time) in each medicinal chemistrylaboratory. This approach has the advan-tage of having a CADD expert sittingalongside medicinal chemists , who canoffer advice on a day-to-day basis, giveinstruction on using CADD tools anddiscuss results and possible next steps ina timely way. This approach fits neatly withthe current trend of creating within theresearch divisions of big Pharma, smallgroupings of researchers operating in a biotech environment.