KNIME Workflows

"ADME properties are calculated with QikProp. The workflow is exported as Computational model in LiveDesign."

Calculates an estimated passive membrane permeability for a ligand structure by comparing the energy optimized conformations of that ligand in low and high dielectric. 
Calculates the amphipilic moment of positively charged molecules.

 In this workflow P450 Site of Metabolism calculation can be set and run to evaluate a set of ligands for the likely sites of metabolism by 2D6 or 2C9 P450 isoform.

Plots a set of QikProp molecular properties to compare the structures.

Generate QSAR models for a chosen property of a set of compounds, and apply the QSAR nodel to other compounds, with either traditional QSAR methods or DeepChem deep learning methods. [Requires: AutoQSAR]

 Assess the coverage of a database from the distribution of the distance to the nearest neighbor of each molecule in the database. List the three most similar compounds for each compound in the database. [Requires: Canvas]

 Pick diverse molecules from a library and inspect the structures. [Requires: Canvas]

Create all possible MCS groups and list the groups identified. Inspect the compounds in the first group and list the compounds that aren't included in this group. List the MCS groups that contain a compound of interest. Create all the groups with the option limiting each compound to at most one MCS group and present the structures in these groups in a matrix. [Requires: Canvas]

 Find the most similar compounds to a sketched molecule in a database. Screen the same database against several query structures. [Requires: Canvas]

 Search a set of structures against a sketched query molecule or SMARTS patterns. Report the compounds that don't pass all the REOS filters. [Requires: Canvas]

Identify and characterize possible ligand binding sites (allosteric sites). Generate the ligand Interaction Diagrams. [Requires: SiteMap]
The SiteMap node is used in more complex workflow examples:
- Real work examples - SiteMap and docking
- Real work examples - Binding site shape clustering"

Compounds are docked with a core constraint using the MCS with the reference ligand.

Dock and covalently binds ligands to the specified reactive residue in the protein target using Michael addition reaction type. [Requires: ProteinPrep, LigPrep]

Dock the ligands in the most similar protein of a set of structures, based on the comparison with the cocrystallized ligand in each of them.

 Dock ligands into multiple conformations of the binding site and report the best pose per ligand. [Requires: Glide]

Align the binding site, and prepare PDB structures, generate the Glide grids and write them to disk. Analyze the protein-ligand complex composition and display the ligand structures in 2D. [Requires: Glide]

 Dock the same set of ligands with several sets of docking parameters and compare the number of top scoring poses. [Requires: Glide]

 Prepare and concatenate a set of co-crystallized, known active, and inactive compounds. Dock them with Glide SP and XP to compare the results using an enrichment plot and a ROC curve. Do the same comparisons after post-processing the results with Prime MM-GBSA. [Requires: Glide, LigPrep, Prime]

 Structure-based Virtual Screening Workflow incorporating ligand preparation, property filtering, and successive docking using different Glide modes (HTVS, SP, XP). [Requires: Glide, QikProp, LigPrep]

 Ensure unique molecule titles (using the KNIME RowID or the Schrodinger dedicated node). This is useful in the context of Canvas. [Version: 2.1][Requires: Maestro, Canvas]

Align a set of ligand structures flexibly with torsional variation, using a best practices approach to choose the most appropriate method. The ligands can be similar (a congeneric series, for example, or multiple congeneric series with different cores) or diverse.

 Illustrate the Output column structure options (output only, input plus output, and output replaces input). The workflow also demonstrates a potential workaround for nodes that don't have the output column structure options yet. [Requires: MacroModel, Glide]

Bioisostere and R-group enumeration using custom or predefined isosteres and libraries. Reaction-Based Enumeration to rapidly generate idea molecules with significant diversity and high probability of being synthesizable.

Create libraries by substituting one or several attachments on a core structure with fragments from reagent compounds. Unique identifiers are assigned to each new compound. [Requires: Maestro]

 Enumerate structures by substitution of R groups in the reactants and products of a single- or multi-step reaction. Also used for Core hopping. [Requires: -]

Usage examples of the nodes operating on a list of models or protocols to upload, download and verify them.
- Create backup copies of some models before they are updated
- Create a model or protocol copy for testing changes and only visible from a specific project
- Copy from a testing to a production LiveDesigninstance
- Deploy the latest version of the validated models available from the KNIME hub
- Deploy the latest version of the generic KNIME protocols
- Compare 2 versions of a protocol or a model

  Structures and properties are retrieved from LiveDesign. New properties are added to an existing Live report.

PDB structures containing ligands similar to each LiveReport structures are listed. The PDB structure with the most similar ligand is prepared and added to a new column in the LiveReport. New columns are also populated with the Ligprepped ligand and its Epik protonation and tautomeric forms.

Calculates ADME properties with QikProp.
Input: Compound structure column
Output column(s): 
- a selection of Qikprop properties

It detects atropisomers in input structures. Each rotatable bond is sampled, and energies are calculated. If the energy difference between the lowest and the highest energy states is sufficient and the energy difference between the two lowest energy states is small, then this bond is assumed to be atropisomeric.
Compound stucture as input column
Output column: 
- Label for structures that may be atropisomers

Adds 2 simple columns to the LiveReport.

Add the binding pocket surface

Minimizes the binding pose conformation to a local minimum and runs a conformational search to find the global minimum.
Calculates the conformation energy difference between the local and the global minima

Replaces functional groups of each input ligand with bioisosteres, generating all the single replacements.

Prepares the protein-ligand complexes in the selected column from the LiveReport and extracts the ligand structures and list.

Docks the compounds with Glide and changes the rendering of the binding site (atoms displayed, residues labelled and the atom representation) to highlight important residues.
The workflow is uploaded as Computational model in LiveDesign.

Docks the compounds with Glide, minimizes the binding pose conformation to a local minimum and runs a conformational search to find the global minimum. Calculates the conformation energy difference between the local and the global minima. Aligns the global minimum conformation on the initial binding pose.

Docks the compounds with Glide. Also generates the protein surface and changes the rendering of the binding site (atoms displayed, residues labelled and the atom representation) to highlight important residues.

Docks the compounds with Glide and changes the rendering of the binding site (atoms displayed, residues labelled and the atom representation) to highlight important residues

Docks the compounds with Glide in 2 binding site conformations and reports the binding site conformation where it scores the best and the corresponding docking pose.

Runs a semiempirical calculation to optimize the ligand geometry and generates the ESP surface.

Extracts simulation quality analysis information from a FEP map column.

Downloads the structure for the PDB IDs listed in the selected Free Form Column from the LiveReport, optionally prepare the structures, extracts and creates new columns with the structures and corresponding structural information.

Adds the binding pocket interaction surface generated with SiteMap.

Aligns ligands on a rigid template structure using Shape screening.

Aligns ligands on a reference structure treated flexibly as well, using Shape screening.

Aligns the structures on a reference ligand kept rigid using align_ligands utility.

Input column: Compound structures in the LiveReport
Configuration: The pharmacophore hypothesis file can be changed on LiveDesign admin page or from the Parametrize model panel.
Aligns the ligands against a pharmacophore hypothesis using Phase.
Output column(s) in the LiveReport: 
- 3D ligand alignment on the pharmacophore hypothesis
- Matched sites
- alignment score

Generates radar plots for the CNS desirability MPO properties.

Generates radar plots of the CNS desirability MPO properties and the number of rotatable bonds with the optimal property range.

Generates a radar plot of the QikProp properties selected in the workflow.

List LiveDesign protocols, models and corresponding parametrized models. Consider archiving some models or unarchive them if needed.

Runs a MacroModel or ConfGen conformational search structures prepared with LigPrep

Input column: Compound structures in the LiveReport
Runs another workflow remotely on a KNIME server.

Generates protonation forms with Epik and adds a new column with the structure of the most probable form.

Reports the predictions of a pregenerated random forest QSAR model run on physicochemical properties generated with QikProp from the LiveReport compound structures.
The QSAR file can be changed on the model admin page or from the Parametrize model panel.

PDB structures containing ligands similar to each LiveReport structures are listed. The PDB structure with the most similar ligand is prepared and added to a new column in the LiveReport. New columns are also populated with the Ligprepped ligand and its Epik protonation and tautomeric forms.

Generates descriptors that can be used in a machine learning model to predict bulk properties such as ionic conductivity, band gap, bulk modulus, formation energy etc.

Various binding configurations of a single ligand in a single binding site are ranked.

An overview (predictions along with the ligand 2D structures) of a set of single edge FEP calculations is given as well as analysis details (including errors and conversion checking). So results can be labeled, triaged, filtered, and extra data can be merged.

A Desmond molecular dynamics simulation is run on each ligand prealigned in the protein binding site. Then the representative frames are refined with Prime and the RMSDs are reported to assess the pose stability. 
The workflow can be used to refine for instance on a docking based alignment to prepare the input of a FEP calculation.

 Prepare a system and run a Desmond MD simulation. Then some frames are extracted and analyzed in terms of hydrogen bond network and atom distance. [Requires: Desmond][Keywords: Molecular dynamics simulation, trajectory, snapshots, jobcontrol, distance, hydrogen bond, atom selection]

 Run a conformational search with various methods and compare the lowest energy conformers. Parse the log file to extract relevant information. [Version: 1.7][Requires: MacroModel, ConfGen][Keywords: conformational search, parsing]

Starting from a linear compound macrocyclizations are created, the conformations sampled and compared for a cocrystallized ligand.

 Screen a file against one hypothesis. Prepare a set of ligands and create a Phase database. Screen this database against several hypotheses using two sets of parameters. Compare the results based on the number of hits found and inspect the hits in Maestro. [Version: 2.4.2][Requires: Phase]

 Use a co-crystallized ligand conformation as a query to screen a set of ligands based on the shape. Screen against several queries. Commands to run the workflow in batch. [Version: 4.0][Requires: Shape Screening]

 The antibody structure can be split into Fc, Fab and Fv domains. The node scans proteins for residues that are candidates for mutation and evaluate the stability and affinity of the mutated structures. Additionally, you can gain access to the latest command-line only functionalities of the tool (e.g., to control Prime loop sampling options), run affinity maturation and aggregation calculations that can be used to compare the mutants. [Version: 2.8.0 Requires: BioLuminate]

Several antibody models are built using Biolumianate tool starting from the heavy and light chain sequences.

Sequences can be read from an XLS spreadsheet. Post-translational modifications and protein descriptors are calculated for each model and the latter can be used as input to BioQSPR.

 Run the default Induced Fit Docking protocol or customize the individual stages. The visual representation of the protein-ligand complexes (molecular representation, color, mutations) can be set before displaying the results in Maestro. [Version: 3.1][Requires: Glide, Prime]

Refine all the loops added using 'Fill loops' option in the Protein Preparation node

 Extract one monomer from each of the two multimers. Align the binding sites (via the 'Align Binding Sites' node) or the whole structure (using the 'Protein Structure Alignment' node). [Version: 2.0.5][Requires: Maestro]

Reactive protein residues are listed and compared for a set of structures. They are also highlighted when inspecting the structures.

 Metanode to run the Sequence converter utility. [Requires: -] [Keywords: Protein]

 Split protein multimers by chain ID and align binding sites. [Version: 5.0][Requires: Maestro]

Reports various metrics that indicate the reliability or quality of a protein structure, and in particular, X-ray structures of proteins.

 A Watermap job is submitted and the map inspected in Maestro. It could be run on several complexes in one run.

 Conformers generated with MacroModel are refined with Jaguar. They are then compared in terms of geometry and Boltzmann populations. A KNIME Report designer template is used to present the results. [Requires: MacroModel, Jaguar]

 Run a semi-empirical and a Jaguar optimization job. Display the ligand ESP charges in Maestro. [Version: 7.0][Requires: LigPrep, Jaguar]

 Evaluate the pKa of specified atoms using Jaguar. [5.3][Requires: LigPrep, Jaguar][Keywords: Quantum mechanics, pKa, parameter flow variables]

 Evaluate the pKa of specified atoms using Jaguar. [5.3][Requires: LigPrep, Jaguar][Keywords: Quantum mechanics, pKa, parameter flow variables]

 Conformers generated with MacroModel are refined with Jaguar. They are then compared in terms of geometry and Boltzmann populations. A KNIME Report designer template is used to present the results. [Requires: MacroModel, Jaguar]

It generates a series of jobs and harvests molecular (including Dipole Moment, HOMO/LUMO, isodensity surface contributions, Internal energy, energy components, enthalpy, entropy and Gibbs free energy contributions) and atomic descriptors (including charges, Fukui indices, electrostatic potential, NMR shielding, spin densities).

 Calculate the molecular orbitals and vibrational frequencies. [Version: 6.0] [Requires: LigPrep, Jaguar][Keywords: quantum mechanics, semi-empirical, parameter flow variables, quick form nodes]

 Run semi-empirical NDDO quantum chemical optimization and generate surface properties. [6.0][Requires: LigPrep, Maestro]

 Analyze a set of PDB structures (or a molecular dynamics trajectory) with SiteMap to characterize frames for possible binding sites. Performed volume clustering on each Sitemap result to produce an ensemble of binding sites. Generate receptor grids for ensemble docking with Glide.

See details in:

Exploring Protein Flexibility: Incorporating Structural Ensembles From Crystal Structures and Simulation into Virtual Screening Protocols. Osguthorpe, D. J.; Sherman, W; Hagler, A. T. J. Phys. Chem. B 2012, 116, 6952–6959

Generation of Receptor Structural Ensembles for Virtual Screening Using Binding Site Shape Analysis and Clustering. Osguthorpe, D. J.; Sherman, W; Hagler, A. T. Chem Biol Drug Des 2012; 80: 182–193

[Version: 6.1][Requires: SiteMap, Canvas, Glide][Keywords: binding site volume, clustering, docking, molecular dynamics]

 Label compounds from a single vendor. Prepare and filter compounds by various criteria.

This workflow takes as inputs: a list of PDB structure IDs, corresponding cocrystallized ligand and Matched Molecular Pair smiles. It downloads the biological units, runs the protein-ligand complex preparation and checking. The ligand structures are extracted, the MMP created and the FEP calculation input file generated.

 Prepare a set of PDB structures and identify possible binding sites with Sitemap. Generate receptor grids with Glide for each site and used to dock ligands. [Version: 5.4][Requires: SiteMap, Glide][Keywords: binding site, ensemble docking, grid generation, druggability]

 The Python Script nodes are used to extract ring properties, run a Macromodel conformational search with specific parameters, and measure distances between atoms for the conformer generated. It is also illustrated how to use it to run third party tools. [Version: 2.4.2][Requires: MacroModel][Keywords: Python, scripting]

 The Run maestro command node is used to alter the formal charge of some atoms defined by an ASL and for altering the structure rendering before inspection of the OPLS partial charges. [Requires: Maestro, Macromodel]

Generate QSAR models for a chosen property of a set of compounds, and apply the QSAR nodel to other compounds, with either traditional QSAR methods or DeepChem deep learning methods.

 Run workflows using the simplified batch execution wrapper script.

Runs various test tools and gather information that help diagnosing installation issues or node failures.

 The Setup Diagnosis node reports possible problems in the plugin installation and set-up. KNIME_batch.py script is run to install KNIME features and update Schrodinger extensions in an existing or new KNIME installation.

The Jobcontrol node can be used to list the currently and recently run jobs, display information on these jobs, list and inspect the corresponding files. It can also be used to kill some jobs. It can be used in combination with the Postmortem node to generate diagnosis summary files for problematic calculations.

 Lists the latest modified workflows in the current workspace or lists workflows containing specific nodes to assist in finding and comparing workflows. Applied on the Schrodinger and KNIME.org workflow examples. [Version: 5.9][Keywords: workflows, workspace, nodes, size, date]

 Uses the KNIME-Maestro connector (replacing the Run Maestro 1:1 metanode) to alter some structures in the middle of a workflow, pick residues to sample in a binding site. [Requires: Prime][Keywords: Maestro commands, scripting]

 Automate some rendering preparation and inspect structures in Maestro. Select residues interactively in Maestro, this selection is used to build an ASL. [Version: 3.0.0][Requires: Maestro]

 Display structure in PyMOL. Run PyMOL commands and save the modified structures. Raytrace views and save the images. [Requires: Maestro, PyMOL][Keywords: PyMOL, visualization]

 Illustrate how to iterate over structures in the input table with two different looper implementations. Extract the second lowest energy conformer of each molecule after a conformational search. [Version: 2.5]

 Illustrate how to use the GroupBy node in the context of the ionization form prediction over a range of pH. Various aggregation methods are used: maximum/minimum, unique count, unique concatenate, set, and list. [Version: 2.0]

Parses a text cell into independent rows and columns that can be postprocessed or visualized more easily.

 Use a loop to unpivot the data. [Version: 2.1]

 The Chemistry external tool nodes can be used to parse a log file with basic shell commands, run Schrödinger utilities or backends with specific options. Quick form nodes can be used to create a configuration panel for this node. [Requires: Maestro, (Phase, MacroModel)]

 Clusters conformers based on Cartesian or torsional RMSD. [Version: 2.9.0][Requires: Maestro, (ConfGen)]

 Predicts ligand states with Epik and reports the ionization form matrix. Generates with LigPrep minimized ionization-tautomeric forms ready to be docked. [Version: 2.9.0][Requires: Epik, LigPrep]

Opens the Multiple Sequence Viewer panel to visualize sequences or edit an alignment. [Requires: Maestro]

 Reader and write structure files. Read a list of PDB files from a directory. [Version: 3.0.0][Requires: Maestro]

 Extract the atom-level properties to list the PDB conversion problems. [Version:3.2.0] [Requires: -]

Most commonly used workbench nodes. [Requires: -]