Icon

Evotec_​StructureBasedVirtualScreening_​Calibration

Workflow for Structure-based Virtual Screening calibration

There are two main parts
1) decoy generation
2) docking evaluation using ROC curves and RMSD calculations

[Requires : RDKit, ChemAxon, Gold, MOE]

Scoring evaluation Decoys generation Read and calculate properties for reference compounds. File should have at least two columns: SMILES & ID Steps:1 - read all compounds from the library you want to screen 2 - calculate properties for all the compounds: MW, LogP, HBDon, HBAcc, Rotatable bonds3- calculate formal charge: not included in the fitlering process after --> to choose if we want to choose decoys with same chargeor not and include it into the filtering step filtering step:for each reference compound, compounds with similar propertiesamong the commercially available ones are selected to be usedfor fingerprint calculation and final decoys selection.Default properties and ranges:MW +/-50LogP +/- 0.5HBDon +/-1HBAcc +/-1Rotatable bonds +/-2 Fingerprints:RDkit topological fingerprints caluclated for reference and potential decoysFingerprint similarity: Tanimotorows are sorted by increasing similarity first 10 rows (10 compounds with lowest similarity) are written as decoys, one file per reference active molecule Ligand preparation Plot AUC versus "number of ligands with RMSD <2Angströms" Ability to retrieve cocrystallised binding modes The 3 metanodes below are containing this small WF in green To handle consensus ranking WARNINGWhen active compounds arereally similar, decoys selectedcan also be similar and we canend up with less than 10decoys for one activecompound. We develop thisWF to ensure that we pick 10decoys (you can change thisvalue directly in the rowsplitter) per active. In red you will find tables with data embed in the WF, for RXRalpha test case, structure 1MVC MW LogP HBDon HBAccRotatable bondssmiles to RDKitsmiles to RDKitMW LogP HBDon HBAccRotatable bonds filtering step RDKit FP RDKit FPsort rows by increasingsimilaritykeep top 500 rowscorresponding to mostdissimilar compoundsremove FP columnall decoysCalculate tanimoto similarity between referenceFPand potential decoys FPTo export file with actives and decoysto transform MOE scoresbigger score is, better docking is-colonneProtein file in mol2 with active site residues (MOE format)here 1MVC.Aligands = activestatus 1decoys = inactivestatus 0structure column = MoleculeID column = IDstructure column = MoleculeID column = IDnumber of unspecified stereocentersmax 2-colonne-colonne-colonneOnly Activesstatus=1Put the database(s)/collection(s) you will screenCrystal ReferencesID should be the same compared to docking resultsunique for each activemseqall active poses with mseqall poses by goldscore for PLP for ASP for chemscoreSort for each scoring function5 rescoring functionsPDB& docking functionKeep the filenameFirst RMSDRMSD <=2CHANGE IT if neededCount ScoringFunctionRMSD Below 2Asort by IDmseqscoring functionsAUCunique decoysretain row orderoutput from docking 1To replace RunGold that is Evotec in-house nodeoutput from docking 2 output from docking 4 unique for each activesort by IDthen by scoretransform scores to ranksInclude score columns onlyaverage rankschemscoreLondon_dG & AlphaHB!!! Minimize NOT Rebuild3D10 decoysfor each active compoundthis value can be changedconfigure Last row numberremove decoysspecific to one active compoundkeep only decoys for each active cpdusing iteration numberto resort table at each iterationremove current iterationin the new tableFirst(Iteration) mustbe integeroutput from docking 3 keep only rank columnsstatus column should be as integerReference compounds ZINC 7K (most dissimilar subset of ZINC 100K) Reference compounds : 33 real ligands goldscore + rescoring chemscore + rescoring asp+rescoring plp+rescoring Protein file in mol2 with active site residues (MOE format)here 1MVC.A 33 Crystal ReferencesID should be the same as from docking resultsFolder with docking results for all poses AUC RMSD Below 2A 33 actives330 decoys in 3D for dockingAUCRMSDgoldscorechemscoreASPPLPRDKit DescriptorCalculation RMSD RDKit From Molecule RDKit From Molecule RDKit DescriptorCalculation Java SnippetRow Filter Chunk Loop Start RDKit Fingerprint RDKit Fingerprint Sorter Row Splitter Column Filter Loop End FingerprintSimilarity MolExporter SDF Extractor Complex Scoring Java Snippet GroupBy Mol2 Reader Column Rename Concatenate (Optionalin) (deprecated) GroupBy Column Rename(Regex) Column Rename(Regex) Column Rename SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) EnrichmentPlotter (local) Concatenate String Manipulation String Manipulation Column Rename Column Rename Set MolNameAnd Comment Row Filter Joiner Java Snippet Java Snippet Java Snippet Joiner ROC Curve (local) Row Filter SDF Reader Read MDB GroupBy Counter Generation Column Rename Ungroup Write MDB RMSD_WriteMDB RMSD_WriteMDB RMSD_WriteMDB Tautomers Major Microspecies Column Filter AdvancedMolConverter Java Snippet Extract Variables(Data) Inject Variables(Data) Sorter Table Creator Concatenate Row Filter Extract Table Spec Table Row ToVariable Loop Start Loop End ConstantValue Column URL to FilePath (Variable) GroupBy Row Filter GroupBy ConstantValue Column Column Rename Java Snippet CSV Writer Column Filter RowID Sorter Joiner Column Rename Chunk Loop Start Table Row to Variable(deprecated) Loop End 2D/3D Scatterplot CSV Writer Column Rename Cell Splitter(deprecated) RowID GroupBy SDF Reader SDF Reader SDF Reader Counter Generation Sorter String Manipulation Insert ColumnHeader Table Rowto Variable Column ListLoop Start Rank ExtractColumn Header Loop End (ColumnAppend) Math Formula EnrichmentPlotter (local) Stereoisomers Standardizer Conformers Energy Minimizer Table Rowto Variable Row Splitter RecursiveLoop Start Recursive Loop End Row Filter Cell Splitter Column Rename Column Filter Row Splitter Concatenate Sorter Column Rename Cell Splitter Column Filter String Manipulation Column AutoType Cast Column Rename Column Filter SDF Reader Column Filter Column AutoType Cast SDF Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader List Files Table Reader Table Reader Table Reader Table Reader CSV Reader CSV Reader Read MDB RunGold RunGold RunGold RunGold AdvancedMolConverter AdvancedMolConverter Scoring evaluation Decoys generation Read and calculate properties for reference compounds. File should have at least two columns: SMILES & ID Steps:1 - read all compounds from the library you want to screen 2 - calculate properties for all the compounds: MW, LogP, HBDon, HBAcc, Rotatable bonds3- calculate formal charge: not included in the fitlering process after --> to choose if we want to choose decoys with same chargeor not and include it into the filtering step filtering step:for each reference compound, compounds with similar propertiesamong the commercially available ones are selected to be usedfor fingerprint calculation and final decoys selection.Default properties and ranges:MW +/-50LogP +/- 0.5HBDon +/-1HBAcc +/-1Rotatable bonds +/-2 Fingerprints:RDkit topological fingerprints caluclated for reference and potential decoysFingerprint similarity: Tanimotorows are sorted by increasing similarity first 10 rows (10 compounds with lowest similarity) are written as decoys, one file per reference active molecule Ligand preparation Plot AUC versus "number of ligands with RMSD <2Angströms" Ability to retrieve cocrystallised binding modes The 3 metanodes below are containing this small WF in green To handle consensus ranking WARNINGWhen active compounds arereally similar, decoys selectedcan also be similar and we canend up with less than 10decoys for one activecompound. We develop thisWF to ensure that we pick 10decoys (you can change thisvalue directly in the rowsplitter) per active. In red you will find tables with data embed in the WF, for RXRalpha test case, structure 1MVC MW LogP HBDon HBAccRotatable bondssmiles to RDKitsmiles to RDKitMW LogP HBDon HBAccRotatable bonds filtering step RDKit FP RDKit FPsort rows by increasingsimilaritykeep top 500 rowscorresponding to mostdissimilar compoundsremove FP columnall decoysCalculate tanimoto similarity between referenceFPand potential decoys FPTo export file with actives and decoysto transform MOE scoresbigger score is, better docking is-colonneProtein file in mol2 with active site residues (MOE format)here 1MVC.Aligands = activestatus 1decoys = inactivestatus 0structure column = MoleculeID column = IDstructure column = MoleculeID column = IDnumber of unspecified stereocentersmax 2-colonne-colonne-colonneOnly Activesstatus=1Put the database(s)/collection(s) you will screenCrystal ReferencesID should be the same compared to docking resultsunique for each activemseqall active poses with mseqall poses by goldscore for PLP for ASP for chemscoreSort for each scoring function5 rescoring functionsPDB& docking functionKeep the filenameFirst RMSDRMSD <=2CHANGE IT if neededCount ScoringFunctionRMSD Below 2Asort by IDmseqscoring functionsAUCunique decoysretain row orderoutput from docking 1To replace RunGold that is Evotec in-house nodeoutput from docking 2 output from docking 4 unique for each activesort by IDthen by scoretransform scores to ranksInclude score columns onlyaverage rankschemscoreLondon_dG & AlphaHB!!! Minimize NOT Rebuild3D10 decoysfor each active compoundthis value can be changedconfigure Last row numberremove decoysspecific to one active compoundkeep only decoys for each active cpdusing iteration numberto resort table at each iterationremove current iterationin the new tableFirst(Iteration) mustbe integeroutput from docking 3 keep only rank columnsstatus column should be as integerReference compounds ZINC 7K (most dissimilar subset of ZINC 100K) Reference compounds : 33 real ligands goldscore + rescoring chemscore + rescoring asp+rescoring plp+rescoring Protein file in mol2 with active site residues (MOE format)here 1MVC.A 33 Crystal ReferencesID should be the same as from docking resultsFolder with docking results for all poses AUC RMSD Below 2A 33 actives330 decoys in 3D for dockingAUCRMSDgoldscorechemscoreASPPLPRDKit DescriptorCalculation RMSD RDKit From Molecule RDKit From Molecule RDKit DescriptorCalculation Java SnippetRow Filter Chunk Loop Start RDKit Fingerprint RDKit Fingerprint Sorter Row Splitter Column Filter Loop End FingerprintSimilarity MolExporter SDF Extractor Complex Scoring Java Snippet GroupBy Mol2 Reader Column Rename Concatenate (Optionalin) (deprecated) GroupBy Column Rename(Regex) Column Rename(Regex) Column Rename SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) SDF Extractor GroupBy Complex Scoring Column Rename Column Rename(Regex) EnrichmentPlotter (local) Concatenate String Manipulation String Manipulation Column Rename Column Rename Set MolNameAnd Comment Row Filter Joiner Java Snippet Java Snippet Java Snippet Joiner ROC Curve (local) Row Filter SDF Reader Read MDB GroupBy Counter Generation Column Rename Ungroup Write MDB RMSD_WriteMDB RMSD_WriteMDB RMSD_WriteMDB Tautomers Major Microspecies Column Filter AdvancedMolConverter Java Snippet Extract Variables(Data) Inject Variables(Data) Sorter Table Creator Concatenate Row Filter Extract Table Spec Table Row ToVariable Loop Start Loop End ConstantValue Column URL to FilePath (Variable) GroupBy Row Filter GroupBy ConstantValue Column Column Rename Java Snippet CSV Writer Column Filter RowID Sorter Joiner Column Rename Chunk Loop Start Table Row to Variable(deprecated) Loop End 2D/3D Scatterplot CSV Writer Column Rename Cell Splitter(deprecated) RowID GroupBy SDF Reader SDF Reader SDF Reader Counter Generation Sorter String Manipulation Insert ColumnHeader Table Rowto Variable Column ListLoop Start Rank ExtractColumn Header Loop End (ColumnAppend) Math Formula EnrichmentPlotter (local) Stereoisomers Standardizer Conformers Energy Minimizer Table Rowto Variable Row Splitter RecursiveLoop Start Recursive Loop End Row Filter Cell Splitter Column Rename Column Filter Row Splitter Concatenate Sorter Column Rename Cell Splitter Column Filter String Manipulation Column AutoType Cast Column Rename Column Filter SDF Reader Column Filter Column AutoType Cast SDF Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader Table Reader List Files Table Reader Table Reader Table Reader Table Reader CSV Reader CSV Reader Read MDB RunGold RunGold RunGold RunGold AdvancedMolConverter AdvancedMolConverter

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: epihan
Created at: 2019-06-25
On NodePit since: 2021-09-28
Last update: 2025-01-20
Created with KNIME version: v4.2.4
Tags: decoysbvscalibrationdockingvirtual screeningstructure-based virtual screeningtuningRMSDROC

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!