Many Schrödinger customers know Dr. Eyrich as the Prime project leader. In addition to his involvement with Prime and his role as Director of Strategic Software Development, Dr. Eyrich has also overseen the creation of Schrödinger KNIME Extensions, which is now available in beta. Here, Dr. Eyrich discusses the functionality users will see in this first release of Schrödinger KNIME Extensions.

We are excited to announce the recent beta release of Schrödinger KNIME Extensions, which allow researchers to rapidly build and execute complex scientific workflows using the KNIME interface. The Schrödinger nodes build upon the existing KNIME infrastructure that already contains a variety of useful nodes, ranging from database reading/writing, statistical analysis, data mining, model building, reporting, and more. With close to 100 nodes covering many core functionalities within the Schrödinger Suite, as well as pre-packaged workflows that incorporate these nodes, we are confident that the first release of KNIME provides a solid foundation for researchers needing to rapidly prototype, validate, and deploy robust computational workflows.

KNIME has established itself as the leading open-source data pipelining tool, and provides an ideal platform for researchers looking for a way to combine best-of-breed technologies, both commercial and academic. KNIME is developed at the University of Konstanz, Germany by the Chair for Bioinformatics and Information Mining. Under the direction of Dr. Michael Berthold, who uses KNIME extensively for research, a large number of new data analysis methods developed at the University have been integrated in KNIME.

Independent of the Schrödinger Extensions, KNIME already includes over 100 processing nodes for data manipulation and mining. The KNIME distribution also incorporates the complete set of analysis models from the well-known Weka data-mining environment and includes plug-ins that allow R-scripts to be run, which gives users access to a vast library of statistical routines. Visualization of results is supported by means of KNIME nodes that support interactive use of scatter plots, parallel coordinates, and more. KNIME is based on the Eclipse platform, providing a modular API that is easily extensible.

Additionally, custom nodes and data types can be developed and integrated into KNIME with a very moderate amount of effort, thus enabling the incorporation of in-house applications and academic packages. Depending on the complexity of the node or data type, this can be done in hours or even minutes.

With Schrödinger KNIME Extensions the number of available nodes is almost doubled, bringing a wealth of ligand- and structure-based chemistry functionality. The current beta release of the Schrödinger KNIME Extensions includes nodes for programs like Glide, Prime, Phase, MacroModel, and Jaguar. This allows for a wide array of calculations to be performed, ranging from rapid conformational searches to more detailed molecular mechanics and quantum mechanics calculations. Complex workflows can be constructed to bring molecules through a series of different programs that compute energetic and structural quantities, which can then easily be combined to build models and improve the accuracy of predictions. Furthermore, a set of key functionalities from our upcoming cheminformatics product, Canvas, allows for many types of fingerprints to be computed, similarity/diversity analysis, clustering, and more.

Schrödinger KNIME nodes support practical considerations as well. Nodes that incorporate LigPrep and Epik functionality can convert molecules from 1D or 2D virtual libraries into 3D structures suitable for both ligand- and structure-based studies. Many nodes exist for protein preparation as well, including the protein assignment code to optimize hydrogen bond networks. A number of fragment-based nodes allow users to perform rule-based fragmentation of existing molecules, or join fragments into molecules. The QikProp node computes ADME properties that can then be used for filtering, data analysis, and model building. Additionally, a variety of visualization tools are available to better explore molecules of interest and associated data.

The above is just a brief listing of what we already have available. Most importantly, we are working hard to bring much more functionality soon, helping to make KNIME the premier modular data exploration tool.

Schrödinger provides support for both the Schrödinger KNIME Extensions as well as for the KNIME platform itself. The KNIME developers also provide support through a KNIME community that contains useful information and discussion forums. We hope to see many other groups, both academic and commercial alike, create nodes to expose the functionality of their programs to the KNIME community.

KNIME can be downloaded with the recently released update of the Schrödinger 2007 Suite. Take a look at what we have and let us know what you think or how you envision using it in your everyday work. Your feedback is always appreciated and helps us expand the product’s capabilities in the directions that are most useful for your research.

At top, the KNIME interface containing a workflow that clusters molecules based on calculated fingerprints. One of the end results of the workflow is the interactive table populated with representative structures that were output from the clustering job. Below, a close-up on several nodes of a KNIME workflow that uses Schrödinger Extensions. The drag-and-drop interface of the KNIME platform makes it relatively easy and intuitive to create new workflows.