KNIME

desktop and Schrödinger extensions overview

Jean-Christophe Mozziconacci and Volker Eyrich

h#p://www.schrodinger.com/upload/KNIME_Overview.pdf
 Schrödinger extensions
•  150+ nodes:
–  Covering most of the Schrödinger tools
Most recent addiGons: Residue scanning, Prime Energy, SiteMap, PyMOL, Glide grid writer
Many take in and output sdf and pdb or mol2 on top of Maestro format
Newest nodes use the same configuraGon panel as Maestro (see the Residue scanning node)
–  Structure and data manipulaGon nodes (eg Split by structure, delete atoms)
–  GUI nodes (eg Run Maestro, Run PyMOL)
–  ScripGng nodes: Run Maestro command, Chemistry external tool, Python nodes
–  UGlity nodes: Setup diagnosis and workflow list
•  50+ workflow examples
–  Workflow page: descripGons and download the ones of interest
–  Whole set can be downloaded with the Suite
–  Many other workflow draXs available on demand
 Schrödinger extensions
•  Tested with latest KNIME version and include the version available at the Gme of
the release
•  Parameter flow variable capability
–  Use the opGons not exposed in the configuraGon panels (eg command line only)
–  Implemented for the main nodes eg Glide, MMod (eg OPLS 2.1), Jaguar
•  2D renderer
–  2D coordinate generaGon and rendering
The default renderer can be set for Maestro columns
–  Can be used in the Report designer
 Schrödinger extensions
•  More and simplified start-up opGons
and stand-alone installaGon configuraGon in the Preferences
–  In $SCHRODINGER/knime start-up script
–  eg use a stand-alone KNIME installaGon, set temporary directories, the memory limit…
•  KNIME menu in Maestro
–  Connect to KNIME mode to exchange structures with a KNIME session
KNIME-Maestro connector node (Improved in 2015-1: automaGc connecGon/close, more modes)
–  Build, import, edit and run workflows from Maestro on project table data
Dynamically generated GUI to alter some parameters.
•  Simplified batch execuGon: KNIME_batch.py
–  Batch command generaGon based on workflow annotaGons
Stand-alone dynamically generated GUI
–  Useful opGons eg -stderr/out
•  Easier installaGon creaGon and update
–  Using Eclipse machinery in KNIME_install.py
–  eg list of extensions to install, from several (zipped) update sites
 Schrödinger extensions
See details in:
•  h#p://www.schrodinger.com/upload/KNIME_Overview.pdf
•  The new features slides:
–  Parameter flow variables (2014-1)
–  Simplified batch execuGon (2013, 2014-1)
–  Chemistry external tool node improvements
–  KNIME menu in Maestro (2012)
–  Start-up script (2012)
•  Workflow examples:
–  Labs > Parameter flow variable usage
–  General > InstallaGon (KNIME_install.py scenarios)
–  General > Workflow list
–  General > Chemistry external tool node usage
–  Phase > Shape screening, docking (batch execuGon)
–  2D renderer in the report designer
 Why use KNIME
Automation Combine various Tools
-  Automate repetitive tasks (especially complex / -  Schrodinger and third party tools (vendor agnosGc),
error prone tasks) scripts (Python, Java, shell),
-  e.g. PDB set preparation, Ligand database -  e.g. QM workflows

preparation
ReporDng
CollaboraDon -  Nodes for reporGng results
-  Share workflows with colleagues (also in Maestro and -  e.g. Enrichment plo#er node, pdf, tables, etc.
Seurat) -  KNIME report designer (free), web portal
-  Not a black box (KNIME.com)

-  e.g. Ensemble docking, HTS analysis
Inexpensive
Prototyping
-  Test parameters
-  e.g. Validate docking parameters

DocumentaDon
-  Easy debugging, interrupGons, data included, inspect
each step (2D, in Maestro/PyMOL)

 Overview
•  Organized by level:
–  Get started
–  Intermediate
–  Advanced funcGonaliGes
•  And by topics:
–  KNIME desktop: GUI, specificiGes, nodes
–  Schrödinger extensions: specificiGes, nodes
•  You can jump between the secGons using links (marked with ►
or◄). See the overview slides.
•  There are also links to use-case examples (marked with ♦).
 Get started
KNIME desktop
– GUI ►
– SpecificiGes ►
– Nodes ►
Schrödinger extensions
– SpecificiGes ►
– Schrödinger nodes ►

► Intermediate

 KNIME desktop GUI ◄
•  Knime.org and Knime.com
•  KNIME desktop
•  Start Knime
•  Create a new workflow and organize a workspace
•  Run a node
•  Import and export workflows
•  Tips and tricks
•  DocumentaGon
 Konstanz InformaDon Miner and Ecosystem
KNIME.org
-  Leading open-source ‘pipelining/workflow’ tool
-  Freely available to academic and industrial researchers
-  KNIME Desktop, based on Eclipse h#p://www.knime.org
-  Community contributions:
-  Modeling tools
-  Marvin sketcher
-  RDKit
-  Indigo
-  CDK
-  R Scripting
-  Erlwood
-  Image Processing
-  HCS Tools
-  Next Generation Sequencing
-  Palladian (mainly GPL3)

KNIME.com Enterprise products and Services h#p://www.knime.com

-  Report designer (free)
-  KNIME team space (share workflows) and KNIME server (web portal and SOA access)
-  Cluster execution (scalability)
-  KNIME professional (support, maintenance and training)
-  Commercial development
 KNIME Extensions
15+ Extension Providers
Extensible, cross-platform, vendor neutral:
Schrödinger, CCG, Tripos, ChemAxon/Infocom, BioSolveIT, Cresset, Dotmatics, Molecular
Discovery, Molegro…

Schrödinger Extensions
-  First released in 2007
-  150+ nodes
-  Molecular mechanics
-  Molecular dynamics
-  Quantum mechanics
-  Cheminformatics
-  Pharmacophore modeling
-  Combinatorial libraries
-  Docking
-  Protein structure prediction
-  Structure and data manipulation
-  Maestro integration
-  Workflow execution
-  Structure exchange
KNIME Desktop GUI
•  Full screen mode
•  Forget about Eclipse
specific menu items

More about:
•  The console ►
 Start KNIME
•  Start up KNIME:
–  On Linux: run $SCHRODINGER/knime
–  On Windows: click on the icon
–  Use -data MyWorkspace to open a specific workspace
–  File > Switch workspace, but KNIME takes Gme to start up again
•  Workspace, workflows and workflow groups:
Create a new workflow and organize a workspace
•  Under the pop-up menu of Workflow Project repository:
–  New KNIME workflow and New Workflow group
–  Copy, Paste, Delete, Move, Rename
•  Drag and drop the workflows in the Workflow project repository
 Run a node
•  Connectors
double click or
•  Node status pop-up

•  Input and output

data tables
 Import and export workflows
•  File > Import KNIME workflow / Export KNIME workflow
or under the pop-up menu of a Workflow Project group
•  Import from another workspace or an archive file (zip)
•  Select 1 or several workflows > export as a zip file
Exclude or not the cached data from the exported file
 Tips and tricks
•  Save regularly the changes. Since KNIME 2.10 there is an auto
save funcGonality but it isn’t on by default.
Save while running calculaGons (see Preferences ►)
•  MulGple undo and redo apply on workflow ediGon (execuGon
data lost though)
•  The KNIME desktop isn’t based on a client-server architecture. If
you close the KNIME instance while a calculaGon is running it
won’t kill it nor you will be able to recover the results when
opening the workflow again.
 DocumentaDon
•  Node descripGons. Also accessible from Help > Help content > Knime > Node descripGons
Type in search field and inspect the configuraGon panel
•  Product page h#p://www.schrodinger.com/KNIME-Extensions
New Features, New Features Slides, KNIME Overview
•  Schrödinger KNIME manual $SCHRODINGER/docs/knime/user_manual/kni13_user_manual.pdf
•  Schrödinger FAQs h#p://www.schrodinger.com/kb
•  Workflow page (examples) h#p://www.schrodinger.com/knimeworkflows/
•  Schrödinger’s extensions webex h#p://www.schrodinger.com/seminarprior/19/24/
•  KNIME.org
–  Workflow examples (preconfigured server access in KNIME explorer)
–  Screencast h#p://www.knime.com/introducGon/screencasts
–  Forum h#p://www.knime.org/forum/
 KNIME desktop GUI ►
•  Preferences
•  Advanced node funcGonaliGes
•  Errors, warnings and Console informaGon
•  Flow variables and workflow variables
•  Metanodes
•  Memory limit
•  Tips and tricks
 KNIME desktop specificiDes ◄
•  Stepwise execuGon
•  Data table column types and conversion

 Stepwise execuDon
•  Only stepwise execuGon
Ideal to take advantage of Schrödinger’s jobcontrol
infrastructure
•  No predefined execuGon order for non connected branches. Use
the flow variable ports [KNIME 2.3]
•  Data cached at each step
•  Preferences > KNIME > Maximum working threads far all nodes
•  1:1 connecGon between nodes (use the concatenate node to
combine input flows)
 Data table column types and conversion
•  KNIME relies on strict data table column typing
•  Converter nodes:
–  Double to Int (integer), String to number, Number to string
–  Molecule type cast (but no Maestro conversion)
–  Openbabel, CDK to molecule, Molecule to CDK
•  In the Schrödinger extensions:
–  String-to-type
–  Molecule-to-MAE, MAE-to-Pdb, MAE-to-SD, MAE-to-Smiles, MAE-to-
mol2, SD-to-smiles
–  Canvas object converters
 Schrödinger specific cell types
•  Structures: Maestro, Sequence, Alignment
•  Several files: Glide grid, Phase Hypothesis
•  Desmond trajectory
•  Binary formats: Canvas fingerprint and matrix








Readers and writers, converters
 KNIME workbench nodes ◄
•  KNIME workbench nodes
•  Data manipulaGon nodes
•  Data exchange
 KNIME workbench nodes
•  I/O nodes for reading and wriGng data from files and
databases
•  Data manipulaGon nodes for managing the internal
data tables that are used to pass informaGon
between nodes
•  CharGng and plopng tools
•  Loop support, Gme Series, Distance matrix
•  StaGsGcs and data mining nodes (Mining, Weka) such
as clustering, neural networks, decision trees, Lib SVM
R staGsGcal computaGon
•  Basic chemistry-aware nodes (CDK)
very limited, see Schrödinger extension nodes ►
The most oPen used nodes for data manipulaDon
•  Row filter, Row spli#er and Sorter
•  Column filter, Column resorter, Column combiner and Rename
•  Joiner (see also Schrödinger Look up and add column node ►) and Concatenate
(only 2 inputs)

And also:
•  Java snippet, RowID and GroupBy node ►
•  Schrödinger nodes for data manipulaGon ►
 Data exchange
•  As text files: File reader and csv writer nodes
•  In Excel format: xls reader and xls writer nodes
•  Between workflows: table reader and table writer
nodes

•  See also among the Schrödinger nodes:
–  Schrödinger reader and writer nodes
–  CSV reader (read several files)
–  View CSV ►
 KNIME workbench nodes ►
•  KNIME.com Labs nodes
•  ScripGng and run a third party tool
•  Java snippet
•  RowID
•  Group by
•  Miscellaneous nodes: InteracGve table, Math formula, CDK Sketcher
•  Plopng faciliGes
•  Looping funcGonaliGes - Basics
•  Model building nodes
 Schrödinger extensions specificiDes ◄
•  Canvas 2D renderer
•  Grouped structures in a cell
•  Output column structure opGons
•  Jobcontrol tab
Canvas 2D renderer
•  Preferences > KNIME > Preferred
renderer
 Grouped structures in a cell
•  #CTs: number of structures
•  Set of conformaGons, Glide poses, Ligprep forms…
•  Group and ungroup nodes, match opGon
•  Also grouped SD, mol2
 Output column structure opDons
•  Input plus Output, Output replaces Input, Output only
•  Extract MAE properGes, Set MAE properGes and delete MAE
properGes nodes.

CTs, columns KNIME new CTs and/or new columns

Set mae properties

Extract mae
properties Delete mae
properties

Reader CTs: Schrödinger Writer,

New CTs
Structures + properties tools Maestro
Output column structure opDons
Jobcontrol
 Schrödinger extensions specificiDes ►
•  Schrödinger preferences
•  Start-up script options
•  Access to flow variables
 Schrödinger nodes ◄
•  Schrödinger node repository
•  Configuration panel visual coherence
•  Nodes of general use:
–  Readers and converters
–  Run Maestro and Run Maestro command
–  Structure manipulation
–  Data manipulation and viewers
–  Scripting

•  KNIME workflow webpage

 Access to Schrödinger tools via KNIME
•  Run on Linux, Mac and Windows CheminformaDcs Pharmacophore Modeling
Fingerprint Based Tools - Phase Shape
32 and 64 bit
- Fingerprint GeneraGon - Phase DB Query
•  Add our extensions to an exisGng KNIME - Generate Pairwise Matrix - Phase File Query
installaGon using the update site - Generate Pairwise Matrix (2 Inputs) - Phase DB CreaGon
•  150+ nodes covering the whole Schrödinger - Similarity Matrix (from Molecules) - Phase Hypothesis IdenGficaGon
Suite - Dissimilarity SelecGon (from Matrix) Docking and Scoring
- Build Report for Clustering (from Matrix) - Glide Grid GeneraGon
- Hierarchical Clustering (from Matrix) - Glide Ligand Docking
Filters and Mining Tools - Glide MulGple Ligand Docking
Molecular Mechanics - Maximum Common Substructure Search - XP Visualizer
- MacroModel Single Point Energy - Substructure Search Post-processing
- MacroModel MinimizaGon - REOS Filter - Prime MM-GBSA
- MacroModel Coordinate Scan - Structure Filter - Embrace MinimizaGon
- ConfGen Standard UDliDes and Converters - Strain Rescore
- ConfGen - Principal Components - Pose Entropy
- ConformaGonal Search - MulG-dimensional Scaling - Pose Filter
- ConformaGonal Search and Cluster - Combine Fingerprints - Glide Ensemble Merge
- Premin - Concatenate Bitvectors - Glide Merge
- Impref - Convert Fingerprint to Bitvector - Glide Sort Results
- Uffmin - Convert Fingerprint to Table Protein Structure PredicDon
Quantum Mechanics - Convert Matrix to Table
- BLAST
- Jaguar Single Point Energy - Convert Table to FingerPrint
- Prime Build Homology Model
- Jaguar MinimizaGon - Convert Table to Matrix
- Prime Side Chain Sampling
- NMR Shielding Constants - Convert Bitvector to FingerPrint
- Prime MinimizaGon
- Jaguar Charges Modeling
- Bayes ClassificaGon Model Building


 Schrödinger nodes
Workflows Desmond Structure ManipulaDon Readers/Writers
Protein PreparaGon - System builder - Add Hydrogens - CSV Reader
- Protein PreparaGon Wizard - Molecular Dynamics - Delete Atoms - Molecule Reader
- Protein Assignment - Trajectory extract frames and manipulaGon - Split by Structure - SD, PDB, Mol2 Reader nodes
Induced-fit docking - Trajectory reader, CMS reader - MAE Parser - Sequence Reader
- IFD and individual steps ReporDng - Extract MAE ProperGes - Alignment Reader
Ligand PreparaDon - Run Maestro - Delete MAE ProperGes - Fingerprint Reader
- LigPrep - Run Canvas - Set MAE ProperGes - Hypothesis Reader
- Ligprep individual tools (Ionizer, - View CSV (open xls/ooffice) - Set Molecule Title - Glide Grid Reader
Desalter, Neutralizer…) - Text Viewer - Set MAE Index - Glide MulGple Grid Reader
- Epik UDliDes - Variable Based Glide Grid Reader
Property GeneraDon Tools - Get PDB - Molecule Writer
- QikProp Combinatorial Libraries - Align Binding Sites - Sequence Writer
- Molecular Descriptors - CombiGlide Library EnumeraGon - Protein Structure Alignment - Alignment Writer
- Calculate properGes - CombiGlide Reagent PreparaGon - Prime Fix - Hypothesis Writer
Filtering Fragments - RMSD - Fingerprint Writer
- Ligfilter - Fragment Joiner - Assign Bond Orders Converters
- Ligparse - Fragments from Molecules - Unique Title Check - Molecule-to-MAE
- Property Filter (Propfilter) Data ManipulaDon - PDB Name - MAE-to-Pdb, to-SD, to-Smiles and to-Mol2
- Compare Ligands - SD Format Checker - SD-to-Smiles
- Lookup and Add Columns - Generate Smarts - PoseViewer-to-Complexes
ScripDng
- Run Maestro Command - Group MAE - Unique Smiles - Complexes-to-PoseViewers
- Chemistry External Tool 0:1, 1:0, 1:1, - Ungroup MAE - Entropy CalculaGon - String-to-Type
1:2, 2:1 and 2:2 nodes - RRHO Entropy - Hartree-to-kcal/mol Converter
- Python Script 0:1, 1:0, 1:1, 1:2, 2:1 and - Boltzmann PopulaGon - kJ-to-kcal Converter
2:2 nodes - Volume Overlap Matrix



 Visual Coherence – Maestro vs. KNIME
Read in
Grid

Read in View
Ligands Results

Docking

Ligand
Preparation
Filtering
 Nodes of general use - Readers and converters
•  Molecule reader, SD reader... Glide grid reader...
•  Converters (Maestro, mae.gz, SD, sd.gz, mol2, PDB, smiles) including Molecule to
MAE, string to type. Canvas converters (Matrix, Fingerprint, Bitvector from and to
table). SD format checker
•  Pose viewer to complexes and Complexes to PoseViewer
Run maestro command and Run Maestro
 Nodes of general use - Structure manipulaDon
•  Set MAE properGes

•  Extract MAE properGes

 Nodes of general use - Structure manipulaDon
•  Extract and Set MAE properGes
•  Group and Ungroup, Set MAE index
•  MAE parser
•  Split by structure, Delete atoms
•  Compare ligands
•  Unique smiles, Unique Gtle check, Ligfilter,
Align binding sites, RMSD,
Volume overlap matrix...
Nodes of general use - Data manipulaDon and viewers
•  Look up and add columns
•  Run Spreadsheet viewer(OpenOffice/Excel)
•  Table viewer
 KNIME Workflows Available for Download

http://www.schrodinger.com/knimeworkflows
 Other KNIME Workflows
CheminformaDcs •  ESP Charges Labs
•  Cluster by Fingerprint •  Jaguar pKa •  Glide Grid Writer
•  Database Analysis
•  Maximum Common Substructure Search (MCS) •  Quantum Mechanical ProperGes •  Parameter Flow Variable Use-cases
•  Select Diverse Molecules •  Run Maestro 1:1 Use-cases
•  Semi-empirical OpGmizaGon
•  Similarity Search
•  Substructure Search General tools
Library Design •  Chemistry External Tool Use-cases
Docking and Post-Processing
•  Library EnumeraGon •  Ensure Molecule Title Uniqueness
•  Docking and Scoring
•  Ensemble Docking •  Output Column Structure OpGon Philosophy

•  Loop Over Docking Parameters •  Protein Structure Alignment
•  Protein PreparaGon and Glide Grid GeneraGon Protein Modeling •  Python Script Node Use-cases
•  Validate Docking Parameters •  Induced Fit Docking Protocol •  Run Maestro Command Node Use-cases
•  Virtual Screening
•  Model Building •  Run PyMOL
•  SiteMap
•  Split and Align MulGmers

Pharmacophore Modeling •  Webservice
•  Phase Hypothesis IdenGficaGon Workbench •  Workflows in the Current Workspace
•  Phase Screening •  Group By Use-cases
•  Shape Screening http://www.schrodinger.com/knimeworkflows/
•  Group Looper

Molecular Dynamics: •  Unpivot

•  Desmond SimulaGon
Real World Examples
Molecular Mechanics •  Binding Site Shape Clustering
•  Compare ConformaGonal Search Methods •  Sitemap and Glide Grid GeneraGon
•  ConformaGonal Search and Post-Processing •  Vendor Database PreparaGon
Quantum mechanics
•  ConformaGonal Search and QM Refinement
 Schrödinger nodes ►
•  Chemistry tool nodes
•  Python nodes
•  Row iterator loop start
•  Look up and add vs. Joiner node
•  Miscellaneous nodes: Compare ligands, Set molecule Gtle
 Intermediate
KNIME workbench
- GUI ►
- Nodes ►

Schrödinger extensions
- SpecificiGes ►
- Schrödinger nodes ►

◄ Get started
► Advanced funcGonaliGes
 KNIME workbench GUI ◄
•  Preferences
•  Advanced node funcGonaliGes
•  Errors, warnings and Console informaGon
•  Flow variables and workflow variables
•  Metanodes
•  Memory limit
•  Tips and tricks
 Preferences > KNIME
•  Directory for temporary files (See also Schrödinger preferences ►)
•  KNIME GUI- disable the node reset, deleGon and reconnecGon confirmaGon
•  KNIME GUI- Console view log level: recommended to change to INFO.
Example of informaGon provided by Schrödinger nodes ►
 Advanced node funcDonaliDes
•  Hovering over an input connector tells you what the node takes as input (eg
Molecules in Maestro, SMILES or SD format)
•  Hovering over an output connector reports the number of rows and
columns in the output table
•  Comment a workflow: Node pop-up menu > Node name and descripGon
•  Data table > change the renderer
 Errors, warnings and Console informaDon
•  Popup-menu > View Std output/error
•  Warning sign above the node status
when the node completed with potenGal
errors.

•  Console informaGon:
INFO HierarchicalClusteringNodeModel Preparing input file '/tmp/HierarchicalClustering_in_423741.mat'
INFO HierarchicalClusteringNodeModel Finished preparing file Gme 0.35 seconds
INFO HierarchicalClusteringNodeModel 10:42:45 11.17.2009:
Running cmdline[0]='=/usr/local/schro-latest/uDliDes/canvasHC -im HierarchicalClustering_in_4116794508031023741.mat -ot
HierarchicalClustering_in_4116794508031023741.tree -og HierarchicalClustering_in_4116794508031023741.csv -linkage schrodinger -n 123'
INFO HierarchicalClusteringNodeModel Completed Gme 1.626 seconds
INFO HierarchicalClusteringNodeModel Preparing output
INFO HierarchicalClusteringNodeModel Finished preparing output: Gme 0.06 seconds
INFO LocalNodeExecuGonJob Hierarchical Clustering (from Matrix) 0:2:50 End execute (2 secs)
 Flow variables and workflow variables
•  The Flow variables are used pass data between nodes on top of the connecGons.
•  In the flow variable tab or the configuraGon panel for a couple of nodes:

•  Global variables can also be set: with the Java snippet node ►
Or in the Workflow project repository select the workflow and Workflow variables in the
pop-up menu.
See also Schrödinger specificiGes ► and nodes to edit variables ►

 Metanodes
•  To hide the complexity and organize a workflow
•  Chose the number and type of input/output
•  The metanodes open up in new tabs
 Memory limit
•  Check the memory limit: Help > About Knime > InstallaGon details > ConfiguraGon and
search for a line starGng with "eclipse.vmargs=-Xmx" (close to the top).
•  Increase the memory allocated to KNIME:
–  $SCHRODINGER\knime -maxHeap 4096m
–  knime –Xmx4096m (as last opGon in the command line)
–  in $SCHRODINGER\knime-v*\bin\*\knime.ini: change -Xmx1024M into 2048M (or
higher on 64 bit)
•  The error message usually contain "Java heap space“ when there is a KNIME is running
out of memory.
•  Preferences > General > Show heap status and use the garbage collector.
•  Knime and Schrödinger tools (eg Canvas) don't compete for memory.
 Tips and tricks
•  Copy and paste some nodes to a specific place:
Select, copy the nodes (Ctrl+C), right click where you want to paste the nodes and
select Paste in the pop-up menu.
Using Ctrl+V instead the nodes will be pasted a li#le below the original ones.
•  The keyboard shortcuts for items on the menus are listed as usual with the menu
item. In File > Preferences > General > Keys you can view all the key bindings to
commands, modify the bindings, and create your own shortcuts.
•  All the branches can be run at the same Gme using the GUI toolbar Execute all
executable nodes bu#on. See also Cancel all running nodes.
 Known issues
•  If you can´t save the workflow with a Java heap space error try
to disconnect the last node or run the garbage collector.

 KNIME workbench GUI ►
•  Report designer
•  Global variables
•  Batch execuGon
•  Tips and tricks
 KNIME workbench nodes ◄
•  KNIME.com Labs nodes
•  ScripGng and run a third party tool
•  Java snippet use-cases
•  Manipulate the table row IDs using the RowID node
•  AggregaGon using the GroupBy node
•  Miscellaneous nodes: InteracGve table, Math formula, CDK Sketcher
•  Plopng faciliGes
•  Looping funcGonaliGes- Basics
•  Model building nodes
 KNIME.com Labs nodes
•  Pipeline Pilot Connector (other way around?)
•  Web Service client, etc
Specific update site: h#p://labs.knime.org/
 ScripDng and run a third party tool
•  Java snippet ►
•  Jython and Schrödinger Python nodes ►
•  Perl scripGng
•  External tool and Schrödinger Chemistry external tool nodes ►
•  Run Maestro commands ►
 Java snippet use-cases
•  Duplicate numeric or string columns
•  Create a new column from scratch (eg a tag)
•  Combine columns (and flow variables) but use the Combiner node for simple
tasks
eg return "prefix-"+$$FlowVar$$+"_ref_"+$Col1$;
•  Add a row index (see also Set MAE index)

See the corresponding
workflow example.
 Manipulate the table row IDs using the RowID node
•  Use data table column values as row IDs and store Row IDs in
a column. Use-cases:
–  before transposing a data table
–  Set the labels to be used by the Plo#er node
•  Ensure row ID uniqueness
–  eg for Canvas tools before creaGng a matrix)
 AggregaDon using the GroupBy node
Some of the aggregaGon methods:
-  first, last
-  max, min
-  Mean
-  Sum
-  Concatenate
-  (unique) count
-  List
-  Set
 Miscellaneous nodes
•  InteracGve table: Find & Find Next
equivalent to the Schrödinger Text viewer node
that have more funcGonaliGes
•  Math formula
•  CDK Structure sketcher
or Marvin sketch
(free of charge from Infocom)
 Ploing faciliDes
•  Data Views: Plo#er, Histogram…
•  Mining > Scoring: Enrichment plo#er, ROC curve
•  Advanced capabiliGes available in KNIME Report
designer ►
Looping faciliDes- Basics
•  Loop start … Loop end
•  Inject and extract variables
•  TableRow/Column to and from variables
•  Prebuilt protocols

•  Schrödinger node: Row iterator loop start ►

Model building nodes

+ Future Canvas nodes

(already some prototypes)
 KNIME workbench nodes ►
•  Edit variables and advanced looping funcGonaliGes
•  Hilite funcGonaliGes
•  Database nodes
•  Miscellaneous useful nodes
 Schrödinger extensions specificiDes ◄
•  Schrödinger preferences
•  Start-up script options
•  Access to the flow variables
 Schrödinger Preferences
•  A specific scratch directory can be specified for Schrödinger nodes
•  Delete temporary files aXer a node successfully executes
Toggled off to run the calculaGon through the command line again

 Start-up script opDons
To pass user/machine/OS-specific parameters

-maxHeap Maximum heap size eg 2048 for 2G
-maxThreads Maximum working threads
-tempDir Schrödinger extensions temporary directory
-defaultHost Default host
-deleteTempFiles true/false Delete temporary files on or off.
-ooCmd <value> Excel / Open Office Spreadsheet command
eg oocalc
or C:\Program Files (x86)\MicrosoX Office\Office12\EXCEL.EXE
And more (see knime –h message)
 Access to the flow variables
•  Access to the flow variables in the Flow variable tab
including the Chemistry external tools nodes (using %flow_n%)
•  Variable based Glide grid reader
 Schrödinger nodes ◄
•  Chemistry tool nodes
•  Python nodes
•  Row iterator loop start
•  Look up and add vs. Joiner node
•  Miscellaneous nodes: Compare ligands, Set molecule Gtle, Get PDB
Chemistry external tool nodes
•  Input/output types:
–  Maestro, SD, mol2,
Smiles
–  Double, Integer
–  String, Text
–  Sequence, alignment
–  FingerPrint, Canvas
Matrix
–  Phase Hypothesis, Glide
Grid

Python nodes
•  Input/output
connectors: 0:1, 1:1,
1:2, 2:2
•  Schrödinger’s APIs
•  Possibility to include
third party APIs
Row iterator loop start
 Look up and add vs. Joiner node
•  Take advantage only 1 column, concatenate columns
The Joiner node is the easiest way to concatenate columns when
the table have the same number of rows and same rowIDs.

 Miscellaneous nodes
•  Get PDB: easy way to get one or several structures
use the | symbol as a delimiter for the list of codes
•  Set molecule Gtle
•  Compare ligands: the modes are First only, both, either
 Schrödinger nodes ►
•  Simple workflow examples
•  Workflow development support for customers
•  AutomaGc protein preparaGon
•  ScienGfically relevant applicaGon of the workflow examples
•  InteracGve work with Knime using the HiLite funcGonaliGes
•  Use a workflow again

 Advanced func+onali+es
KNIME workbench
- GUI ►
- Nodes ►

Schrödinger extensions
- Advanced workflow examples ►

◄ Intermediate
 KNIME workbench GUI ◄
•  Report designer
•  Global variables
•  Batch execuGon
•  Tips and tricks
 Report designer
•  From knime.com but free of charge. Included in our distribuGon
•  Include To report node(s) in the workflow (can’t be in metanodes) and switch to the
Report designer mode
Report designer- template mode
 Report designer- Canvas 2D renderer
The structures can be shown in a report using Canvas 2D renderer using the following procedure:
•  1. In the workflow, add a MAE-to-smiles node and a To report node.
•  2. In ReporGng mode, in the Layout tab, add a table to the report (drag and drop from the Data set view).
•  3. Insert in the "[smiles]" cell (Table- detail row) an Image widget from the Report Items list.
•  4. Configure the widget (using "Edit" on the widget), select "Dynamic image", and press "Select Image
Data..." to select the source column (which should be the Smiles column). Delete "[smiles]" if you want just
the image and no SMILES. You may want to alter the size of the cell by dragging the border verGcally and
horizontally if necessary.
•  5. Change the size of the image to something like 300x300, which is done by ediGng the Data set view
(right click -> Edit -> Parameters), and changing (or creaGng new Parameters typed as integer if they don't
exist yet) the knime-image-height and knime-image-width parameters.
•  6. Check the view in the Preview tab
 Global variables
•  In the Workflow project list, right-click on the workflow, under Workflow variables
 Batch execuDon
•  $SCHRODINGER/knime -batch –reset –nosplash -nosave
-workflowFile=<path>/<wkf>.zip or -workflowDir=<path>/<workspace>/<wkf>
•  Alter some sepngs -opDon=nodeNumber,valueName,value,type
-opGon=7,filename,"/tmp/new-molprops.csv",String (int, double or String)
Find the node number in the configuraGon panel header (add the metanode numbers)
eg 123/456/78 for the node 78 in the metanode 456 in the metanode 123
Find the opGon name in the workspace directory: <workflow>/node_name(#7)/node.xml eg:
<config key="DataURL">
<entry key="array-size" type="xint" value="1" />
<entry key="0" type="xstring" value="/C:/serotonin_unique.sdf" />
-opGon=2,DataURL\0,"file:/tmp/new-input.mae",String
When the input is an array
•  Pass some variables: -workflow.variable=name,value,type (int, double or String)
•  Workflows can be run from Maestro using a simple Python script wrapper
 Tips and tricks
•  Rearrange the panels
•  Workflow Meta-Infos
•  Try to open a workflow modified with a newer version of KNIME alter the 2
following lines of the file
<workspace>/<workflow>/workflow.knime:
<entry key="created_by" type="xstring" value="2.0.3.0021120"/>
<entry key="version" type="xstring" value="2.0.0"/>
 KNIME workbench nodes ◄
•  Edit variables and advanced looping funcGonaliGes
•  Hilite funcGonaliGes
•  Database nodes
•  Miscellaneous useful nodes

Edit variables and advanced looping funcDonaliDes

•  Extract variable (data) and Inject variable (data)

•  TableRow to Variable (use the first row),
Variable to TableRow and Variable to
TableColumn
 Hilite funcDonaliDes
•  HiLite filter and HiLite collector nodes
•  Color, Size and Shape Manager/Appender nodes
 Database nodes
•  See simple examples (not on the Workflow page yet)
Time series support
•  See simple examples (not on the Workflow page yet)
Miscellaneous nodes: Cell spliker, Numeric binner
•  Reference row filter, Reference column filter, Nominal value row filter
•  Missing values
•  Create collecGon column and Split collecGon
•  Text manipulaGon: String replacer, Case converter, Cell spli#er
•  Row sampling, ParGGoning, Shuffle
•  Numeric binner

 KNIME workflow examples ◄
•  KNIME workflow page
•  Workflow development support for customers
•  Workflow example presentaGon
 KNIME workflow page - h#p://www.schrodinger.com/knimeworkflows/
CheminformaDcs
Homology modeling
•  Substructure Search
•  Model building and refinement
•  Clustering, diversity selecGon, similarity search
•  Database analysis Library design
•  Maximum Common Substructure •  Library enumeraGon
Docking and post-processing including a Run Maestro 1:1 prototype
•  Protein preparaGon and Glide grid generaGon Real World Examples
•  Docking and scoring, Virtual screening, Ensemble docking, Induced
•  Vendor database preparaGon
Fit Docking
•  Loop over docking parameters General tools
•  Validate docking parameters •  Python script node use-cases
Pharmacophore modeling including a Run PyMOL prototype
•  Phase Shape screening •  Chemistry external tool node use-cases
•  Phase hypothesis idenGficaGon •  Run maestro command node use-cases
•  Phase database screening •  Output column structure opGon philosophy
Molecular Mechanics KNIME workbench
•  Compare conformaGonal search methods •  Workflows in the current workspace
Quantum mechanics KNIME workbench- looper
•  ConformaGonal search and QM opGmizaGon •  Group Looper
Using the Report designer
Simplest, most exciDng, new and improved workflows
 Workflow development support for customers
Combine or expand the workflow examples
•  Hierarchical clustering and diverse compounds from each cluster
WaiDng for new nodes to be developed
•  CombiGlide library enumeraGon, MacroModel coordinate scan (now available)
•  PCA on per residue interacGons (Chemistry external tool node)
•  Distance measurement in protein and run script in Maestro, descriptor calculaGon, create protein mutants (Python
script node)
Specific workflows
•  Various MacroModel protocols using the Python node for accessing advanced funcGonaliGes (eg contraints)
•  Prime MM-GB/SA on a set of complexes (ligand detecGon, flexible residues)
•  Simplifying compound docking with KNIME, Dr. Robert Happel, Boehringer Ingelheim, Vienna
h#p://www.schrodinger.com/seminarprior/19/26/

•  Cris Guimaraes MM-GB/SA paper reproducGon and improvements

h#p://www.schrodinger.com/Download.php?type=seminarentry&type2=slides&ident=105
•  Protein preparaGon protocol

Real World Case Study: Binding Site Clustering and Ensemble Docking
Real life applicaDons...
 Workflow example presentaDon
Feel free to request this other presentaGon including:
•  Simple examples
•  More advanced examples from the Workflow page
•  ScienGfically relevant applicaGons
 Schrödinger KNIME Extensions
KNIME Extensions Product Manager: Jean-Christophe Mozziconacci
([email protected])
Vice President of Technology: Volker Eyrich ([email protected])
Main Developer: Ravikiran Kuppuraj
and the PyDev development team
QA: Simon Foucher
Technical Support: Katalin Phimister, Pavel Golubkov
MarkeDng: Jarred Yacob