KNIME interfaces

KNIME® interfaces

KNIME® — the Konstanz Information Miner — is a modular, extendable data exploration platform to visually create data pipelines. KNIME® is based on the Java and the Eclipse platform and, through its modular API, easily extensible. When desired, custom nodes and types can be implemented in KNIME® within hours thus extending KNIME® to comprehend and provide first-tier support for highly domain-specific data. This modularity and extensibility permits KNIME® to be employed in commercial production environments as well as teaching and research prototyping settings.

BioSolveIT is a partner of KNIME® and in the process of making its software-tools for drug design and discovery accessible to all users via KNIME®. Note that these interfaces are distributed separately of KNIME® and available free of charge. KNIME® itself is distributed under open source license and can be used freely as long as the software or extensions built into it are not distributed for commercial gain (see details on the KNIME® website).

Small molecule library builder & pocket explorer

BioSolveIT-FragXplorer is a virtual library builder and docking-workflow for drug design and discovery designed to explore binding sites MedChem-style. It supports:
  • evolving a fragment binder (fragment-growing)
  • exploring un-addressed space in a binding site
  • selecting optimal building blocks for purchase
  • designing a small-molecule library tailor-made for the target at hand
Starting from a template small molecule or fragment in a PDB or proprietary x-ray structure, the user first defines a common element of the library (the unchanging core element), and then selects from a pre-installed list the chemical reaction that will create the library (e.g. a particular amide formation). The workflow then docks, scores, and ranks all the members of the virtual library, based on the position of the unchanging core. The molecules predicted to be the most potent are then presented for visual inspection in SeeSAR.

The example shown here is based on a core fragment of sildenafil (in black), a known active against human phosphodiesterase 5A (PDB: 1TBF). FragXplorer outputs multiple interesting ideas for fragment elaboration, of which above only a few virtual products are illustrated. In this example, the Buchwald-Hartwig coupling reaction was chosen to build up the virtual products based on a subset of the emolecules catalog.

Get the FragXplorer workflow here


Finding hits in chemical spaces

With the use of BioSolveIT KNIME® nodes, retrieving molecules from chemical spaces is now very easy. However, this is usually not the end of story but the beginning of a virtual hit triage – leading you to those compounds of biggest interest. The SpaceXplorer-workflow (previously 'REALizer') gives you several post-processing options to quickly get an overview of the best compounds. After searching a Chemical Space you can see:
  • Why FTrees considers your hit molecules similar to your query.
    Based on the unique FTrees-mapping images you will immediately see, how similar parts of two molecules match to each other.
  • A plot of FTrees Similarity versus the well-known Tanimoto Similarity.
    This helps you when you are looking for scaffold hops, because you might find compounds with high FTrees but low Tanimoto similarity.
  • A 3D alignment.
    In case you are working “ligand-based” this workflow will help you identify compounds with high shape similarity to your query.
  • Docking poses.
    If you know the 3D structure of your target the docking-workflow will retrieve from the virtual hits those that fit well into the binding site. Afterwards you may visually inspect the top hits in SeeSAR.
The workflow includes example data, where Sildenafil (PDB: 1TBF) is searched inside the REAL Space. After the search in the chemical space, four alternative ways of post-processing can be executed. Once you try you will see how different results of your search will be top-ranked at the end of the workflow and provide you with new ideas.

Get the SpaceXplorer workflow here


BioSphere - building & searching chemical spaces

Chemical space searching combines the goals of de novo design and synthetic accessibility. The BioSphere workflow provides you with all tools to build and search chemical spaces based on user-defined reactions and reagents. It consists of two parts:
  • The first workflow is to build a space using the brand new CoLibri nodes. Individual reactions can be designed and the virtual products can be merged to an already existing space consisting of over 100 hand selected med-chem reactions altogether constituting the KnowledgeSpace.
  • The second workflow searches in 45 billion (+ the user defined) virtual products. As a user you simply draw a query, which is explored in the space using FTrees-FS. The virtual hits are automatically docked in a user-prepared protein and presented in SeeSAR for visual inspection.
This workflow exemplifies building incredibly huge spaces and searching them within minutes! It gives you access to reaction definitions, which you may use to supplement the space. Filter differently or use your own reagents to customize the space according to your needs. Benefit from the best in class tools to utilize in your research projects.

The workflow includes example data (where a space is built and searched). The reactions used to construct the space can be found in the file 'reactionoverview.pdf'. The compound CIU, a ligand of the soluble Epoxide Hydrolase (sEH, PDB: 1VJ5), was used as a query. sEH is part of the arachidonic acid cascade and plays a role in several diseases, including hypertension, cardiac hypertrophy, arteriosclerosis, brain and heart ischemia, cancer and pain. Some compounds targeting sEH have already reached clinical studies, but no drug is approved yet. Known inhibitors of sEH are ureas like the query molecule CIU. Interestingly, under the results are amides, which are known inhibitors of sEH, too.

Get the BioSphere workflow here


MedChemWizard - exploring the SAR around a compound

The MedChemWizard is a KNIME® workflow designed to assist medicinal chemists with idea generation, ligand design and lead optimization using a number of common functional group transformations and medchem rules-of-thumb. The user simply draws in a hit or lead structure, and the Wizard will enumerate new ideas based on bioisosteric replacements, methylene shuffling, diamine morphing, aromatic ring analogs, aromatic ring substitutions, reversed amides/sulfonamides, new virtual rings, protecting metabolic soft-spots and the replacement of hydrogens with solubilising and other groups.

MedChemWizard integrates with BioSolveIT's SeeSAR KNIME® nodes to provide an estimated binding affinity and a predicted binding mode for generated ligands, allowing the user to interactively inspect the quality of the new ideas and to select those of greatest interest.

It is intended that this workflow will help medicinal chemistry teams avoid spending time synthesizing and testing poor (low potency) compounds, and facilitate scaffold hopping/morphing into new chemical and intellectual property space — together, these factors will help improve overall odds of project success!
The workflow includes example data where the ligand of bromodomain BRD4 (PDB: 6FO5) is modified. The reaction products get filtered and docked.

Get the MedChemWizard workflow here


CovXplorer - covalent drug discovery

The drug discovery workflow CovXplorer was designed for covalent screening campaigns at any target of interest. After definition of the binding site, users can decide if they want to work with preprocessed covalent ligand libraries or use their own in-house structures and libraries. The workflow is able to handle over 30 well-known covalent warheads and can easily be modified to cover other functionalities of interest. The list features tried and tested warheads like:
  • acrylamides
  • activated acetylenes
  • boronic acids
  • and many more
Different filtering options allow users to screen for warheads specifically targeting their residue type or to select a set of warheads to be investigated. The workflow is able to dock compounds non-covalently in the first step to assess the initial binding of the ligand before forming a bond to the target. In a second step interesting candidates are processed into their target-bound form and redocked covalently with subsequent Hyde assessment.

Get the CovXplorer workflow here

BioSolveIT tools in KNIME®
SeeSAR is BioSolveIT's next-generation modeling tool for all drug designers. With an intuitive interface, users can quickly and easily load and visualize protein-ligand complexes alongside an estimated binding affinity as calculated by our unique, proprietary HYDE scoring technology. Additional molecular properties are calculated directly within SeeSAR, whilst custom user properties and descriptors can be included via the SDF Inserter node, allowing the user to interactively sort and filter new ideas. In SeeSAR, ligand atoms are individually depicted not only by atom type, but also by their contribution to binding affinity, allowing the user to rapidly identify potential sites of modification to enhance potency.
Our KNIME® node for predicting small molecule binding affinities (Assess Affinity with HYDE in SeeSAR) allows users to easily access our unique, proprietary HYDE scoring technology non-interactively. Using a pre-prepared binding site, input ligands (SDF or mol2 format) are scored — the outputs are an active-site optimized ligand conformation (SDF) and the predicted binding affinity. These results can easily be sorted by predicted affinity as the node outputs both higher and lower numerical limits of the range.
Performing docking calculations is a routine, yet quite delicate task in every structure-based drug discovery process. We automated the entire process and thereby simplified the UI for you tremendously. All you need to do is provide a PDB and a reference ligand to define a binding site and you are ready to go.
Similarity searching is a task that needs to be performed routinely and oftentimes in conjunction with other tools in a workflow. The FTrees Similarity node compares the molecules in the pipeline to separately provided query molecules. Additionally you may:
  • filter molecules according to similarity
  • generate 2D depictions that visualize the FTrees-mapping using colors
  • compute 3D alignments based on the FTrees-mapping on the fly
Feature Trees Fragment Spaces (FTrees-FS) is a unique technology, which enables the user to perform similarity searches in innumerably large compound spaces. The KNIME®-module comprises the search engine. You may use it for example to perform similarity searches in over 12 billion molecules of the so-called KnowledgeSpace, which based on chemistry protocols to ensure it comprises synthesizable molecules. The KnowledgeSpace can be downloaded free of charge here.
This is the package of choice for generating chemical spaces such as the ones to be searched by FTrees-FS. The package consists of three tools:
  • The Reaction Library Synthesizer takes RXN or SMIRKS as input and builds so-called single-library-spaces (cf. below).
  • The Fragment Space Merger takes single-library-spaces (cf. above) and puts them together in a unified fashion which enables searching with FTrees-FS.
  • The Reaction Library Enumerator works similar to the Synthesizer but enumerates the products of your virtual reaction.
Alignments of small molecules can give much insight into structure activity relationships. This very intuitive interface to our FlexS program allows piping in query and a database to screen.
The generation of trustworthy and realistic 3D coordinates for small organic molecules is of great importance in structure-based drug design and cheminformatics. Our Coord3D tool quickly generates high quality 3D coordinates for small molecules based on gold-standard torsion angle data extracted directly from the Cambridge Crystallographic Data Centre's CSD database. Based on 2D inputs (smiles, SD or mol2 format), the tool will convert and output in 3D in the format of your choice.
The Naomi converter contains a super fast compute-engine that reads and writes compounds in a number of standard file formats. It is also capable of calculating basic molecule properties and to use them for filtering. The engine is able to run in parallel on multi-core machines.

Naomi can convert molecules between different representations (SDF, Mol2, and SMILES) and perform substructure-mappings. In addition it is able to generate high quality 2D images in different graphic file formats (SVG, PNG). These can also be used in KNIME® reports as Naomi can be registered as molecule renderer within KNIME®.

The 2D depictor is based on the latest 2D drawing technology from the Center for Bioinformatics in Hamburg (ZBH).
This is a step-by-step introduction that explains for beginners how to use the BioSolveIT software in KNIME® all the way from the download to a virtual screening workflow executed in parallel:
step-by-step introduction (PDF)

NOTE: In principle any update should seamlessly replace previously installed BioSolveIT KNIME® nodes if these where officially released on our website. However, in some cases this mechanism may fail. Thus, we recommend to first uninstall all prior versions of the BioSolveIT KNIME® nodes before installing the new ones. This can be done very easily if you click on “What is already installed” and follow the instructions.

If you need an older version, please contact us via the support page or email.

  • Only the most recent version of the respective software packages is guaranteed to work with the KNIME® modules
  • The following KNIME extensions are required to install the BioSolveIT KNIME® nodes
    • KNIME Base Chemistry Types and Nodes
    • KNIME SVG Support
  • Supported operating systems: Windows and Linux (partly MacOS)

Install and update BioSolveIT KNIME Interfaces conveniently
by adding this URL to Help → Install New Software within KNIME®:

If you prefer downloading and installing the nodes manually, please select one of these packages:

Get your KNIME Interfaces license: