Lecture 9 molecular descriptors

molecular descriptors

1. Lecture 9-Molecular descriptors
BTT-516–Drug Designing and Development

2. Topic To be covered
1. Introduction
2.Types of molecular descriptors
3. Tools for descriptor calculations
4. Home work

3. Moleculardescriptorscanbedefinedasmathematicalrepresentationsofmolecules’
propertiesthataregeneratedbyalgorithms.
Thenumericalvaluesofmoleculardescriptorsareusedtoquantitativelydescribethe
physicalandchemicalinformationofthemolecules.
AnexampleofmoleculardescriptorsistheLogPwhichisaquantitativerepresentationof
thelipophilicityofthemolecules,itisobtainedbymeasuringthepartitioningofthe
moleculebetweenanaqueousphaseandalipophilicphasewhichconsistsusuallyof
water/n-octanol.
Introduction
Moleculardescriptorscanbeusefulinperformingsimilaritysearchesinmolecular
libraries,astheycanfindmoleculeswithsimilarphysicalorchemicalpropertiesbasedon
theirsimilarityinthedescriptors’values.

4. ThemoleculardescriptorsareusedinADMETpredictionmodelstocorrelatethe
structure–propertyrelationshiptohelpinpredictingtheADMETpropertiesofmolecules
basedontheirdescriptorsvalues(Khanandsylte,2007).
ThemoleculardescriptorsthatareusedinADMETmodelscanbeclassifiedonthebasisof
levelofmolecularrepresentationrequiredforcalculatingthedescriptor.
•One-dimensional(1D)
•Two-dimensional(2D)
•Three-dimensional(3D)
The1Ddescriptorsarethesimplesttypeofmoleculardescriptors,theserepresent
informationthatarecalculatedfromthemolecularformulaofthemolecule,which
includesthecountandtypeofatomsinthemoleculeandthemolecularweight.
One-dimensional(1D)

5. The2Ddescriptorsaremorecomplexthanthe1Ddescriptors,usually,theyrepresent
molecularinformationregardingthesize,shape,andelectronicdistributioninthemolecule.
Calculatingthe2Ddescriptorsdependsmainlyonthedatabasesize,andthecalculationofparts
ofamoleculeinwhichthedataismissingcouldlargelyresultinafalseresult.
The3Ddescriptorsdescribemainlypropertiesthatarerelatedtothe3Dconformationofthe
molecule,suchastheintramolecularhydrogenbonding.
Examplesofdescriptorsobtainedfromcalculationsinvolvingthe3Dstructureofthemolecules
arethepolarandnonpolarsurfacearea(PSAandNPSA,respectively).Moreadvanced
calculationlikequantummechanicscalculationscanbeusedtoobtain3Ddescriptorsthat
describethevalenceelectrondistributioninthemolecules(Bergström,2005).
3D descriptors
2 D descriptors

6. •0D-bondcounts,molweight,atomcounts
•1D-fragmentcounts,H-Bondacc/don,Crippen,PSA,SMARTS
•2D-topologicaldescriptors(Balaban,Randic,Wiener,BCUT,kappa,chi)
•3D-geometricaldescriptors(3DWHIM,3Dautocorrelation,3D-Morse)+surface
properties+COMFA
•4D-3Dcoordinates+conformations(JCHEMconformer,CORINA,goldset,
Crystaleye)

7. Aselectionofcommercialandfreedescriptorcalculationutilitiesiscollectedunderthe
moleculardescriptorsoftwarecollectionortheCompChemlistornewprogramsareposted
toCCL.
•alvaDesc-newvisualdescriptorsuitefromKodesolutionscovering4000descriptors
(developedbyAlvascience)
•CDKdescriptorGUI(freeandopensource-usingOpenSourceCDKandJoelibcode)
•BlueDesc-MolecularDescriptorCalculator(freeandopensource-usingCDKandJoelib
code,requiresJAVA1.6
•ChemAxonJChem-DescriptorpackageusingMarvinJAVAAPI(freeacademiclicense)
•ISIDA/QSPR-freefragmentbasedQSPRdescriptorpackage
•E-Dragon(VCCLab)free(150molecules),nowwithGSFRAG,GSFRAG-L,ETState>
3000descriptors
Toolsfordescriptorcalculations

8. •MOLD2-(FDA)afree2Dmoleculedescriptorpackage
•ToxicityEstimationSoftwareTool(T.E.S.T.)-(EPA)containsmorethan7902-dimensional
descriptors
•Open3DQSAR-pharmacophoremodellingusingmolecularinteractionfields(MIFs)
•Dragon-5,270moleculardescriptorsforLINUXandWIN(Todeschini/Talete/Kode)
•PaDEL-Descriptor-basedonCDKbutincludesadditional7372Dand3Ddescriptors
(NUS/Singapore)
•ADMEWORKS ModelBuilder-400descriptors(Jurs)andMOPAC(Stewart)(Fujitsu/Poland)
•QuBiLS-MIDAS-ahighlyparallelsoftwareforthree-dimensionalmoleculardescriptor
calculation

9. Concepts for descriptor calculations and QSAR/QSPR
modeling
•Youneedalargedatasetwiththemolecularproperty(logP,bp)tobemodeled.The
largerthenumberofdatapointsthebetter.ThereareQSARmodelswith20orless
points,howeverforbroadapplicationsoneneedtocoveralargediversityspace.
Hundredsorthousandsofsuchvaluescanbecollectedfromdatabasesorarenow
availablefromHTscreeningmethods.
•Youneedthemolecularstructuresitself(asSMILES,SDFin2Doroptimized3D
structure).Handlingthemoleculestogetherwithalldescriptorscanbeachallenging
task,softwarewhichcandothatishighlypreferred.
•You need a descriptor package for descriptor calculation
•Youneedtoapplyfeatureselection(astatisticalprocess)todiscardunimportant
(invariant)orsometimeshighlycorrelateddescriptors(othogonalization)

10. •Youneedtodivideyourmoleculesetintothreeparts.Atraining(70%),validation(30%)and
anadditionalexternaltrainingorvalidationsetwhichisnotusedineithermethod.(Sometime
thevalidationsetiscalledtestingsetorviceversa).Cross-validation(n-foldorv-fold)
techniquesorotherresamplingtests(MonteCarloSampling,Jackknifing,Bootstrapping)need
tobeapplied,especiallyifnotenoughmoleculesareavailable.
•Youneedtoapplyregressionorclassificationmethods(includingmeta-learningapproaches).
•Oneneedtomakesurethatforfuturepredictionsnoothercompoundclassesareincluded
(whichusuallyresultsinwrongpredictions)byeitherincludingerrorvalues,fingerprintor
substructurematchesorasimpledimensionreductionmethod(PCA,PLS)toavoidmolecules
whichwerenotcoveredduringdevelopment.AsexamplealogPmethodonlydevelopedon
alkaneswill100%failoncomplexdrugmoleculesormoleculeswithmultiple-OHand-NH
or-SHgroups.Furthermoreacompletestatisticaldescriptionforeithertheregression
performanceorclassificationperformanceneedstobeincluded.

11. Utility of molecular descriptors
•Thepurposeofmolecular-Descriptoristocalculatepropertiesofmolecules
thatserveasnumericaldescriptionsorcharacterizationsofmoleculesin
othercalculationssuchasQSARmodel,diversityanalysisorcombinatorial
librarydesign.

12. Thank you
Er. Rajan Rolta
Faculty of Applied Sciences and Biotechnology
Shoolini University,
Village Bhajol, Solan (H.P)
[phone] (Mob No.)
[email protected]