Icon

03_​Multiobjective Optimization

Read saved models andgenerate predictions 03_Multiobjective OptimizationThis workflow demonstrates how to combine the prediction outcomes from several computational models (activity, clearance, CYP metabolism and hERG binding) to perform multiobjective optimizationof compounds with Pareto ranking. 1. Create a component similar to the others that will allowthe user to define the fingerprint column used to makeactivity predictions. Use the Column Selection Configurationnode, the Model Reader node and the Random ForestPredictor node. A Column Filter node will exclude thefingerprint columns we don't need anymore. Connect the component output to the Joiner node. 2. Use the Pareto Ranking node here and set the followingconfigurations: * Clearance (P(is_low=1)): Minimize* CYP1A2 metabolism (P(cyp1a2_active=1)): Minimize* hERG binding (P(herg_active=1)): Minimize* Activity: Prediction (activity)(Confidence): Maximize* SlogP: Optimal Range 0.0 <> 4.0* NumRotatableBonds: Optimal Range 1.0 <> 4.0* NumHeavyAtoms: Minimize* TPSA: Optimal Range 75 <>140Connect the output to the lower Joiner node . 3. Build an Interactive View to explore the results. Wepropose a Parallel Coordinates plot to visualize the rank, theprobabilities and the molecular properties. Use the ColorManager node to asign some colors based on the Paretorank. Combine it with a Tile view to see the compoundsselected in the Parallel Coordinates plot. Render themolecules as images to display them on the tiles. To remindthe user about the settings of the Pareto Rank, a Text OutputWidget node is helpful. The Row Filter, Column Filter andColumn Resorter node will help you to bring the data inshape for the according view nodes if needed. adjust the file location Clearance model CYP 1A2 Model hERG Model CalculatedProperties Annotate compounds Excel Writer Joiner Joiner Joiner Joiner Table Reader Joiner Read saved models andgenerate predictions 03_Multiobjective OptimizationThis workflow demonstrates how to combine the prediction outcomes from several computational models (activity, clearance, CYP metabolism and hERG binding) to perform multiobjective optimizationof compounds with Pareto ranking. 1. Create a component similar to the others that will allowthe user to define the fingerprint column used to makeactivity predictions. Use the Column Selection Configurationnode, the Model Reader node and the Random ForestPredictor node. A Column Filter node will exclude thefingerprint columns we don't need anymore. Connect the component output to the Joiner node. 2. Use the Pareto Ranking node here and set the followingconfigurations: * Clearance (P(is_low=1)): Minimize* CYP1A2 metabolism (P(cyp1a2_active=1)): Minimize* hERG binding (P(herg_active=1)): Minimize* Activity: Prediction (activity)(Confidence): Maximize* SlogP: Optimal Range 0.0 <> 4.0* NumRotatableBonds: Optimal Range 1.0 <> 4.0* NumHeavyAtoms: Minimize* TPSA: Optimal Range 75 <>140Connect the output to the lower Joiner node . 3. Build an Interactive View to explore the results. Wepropose a Parallel Coordinates plot to visualize the rank, theprobabilities and the molecular properties. Use the ColorManager node to asign some colors based on the Paretorank. Combine it with a Tile view to see the compoundsselected in the Parallel Coordinates plot. Render themolecules as images to display them on the tiles. To remindthe user about the settings of the Pareto Rank, a Text OutputWidget node is helpful. The Row Filter, Column Filter andColumn Resorter node will help you to bring the data inshape for the according view nodes if needed. adjust the file location Clearance model CYP 1A2 Model hERG Model CalculatedProperties Annotate compounds Excel Writer Joiner Joiner Joiner Joiner Table Reader Joiner

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: glandrum
Created at: 2020-04-15
On NodePit since: 2021-09-18
Last update: 2024-04-25
Created with KNIME version: v4.4.1

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

Try NodePit Runner!