KNIME: News

03 June 2008: KNIME 1.3.5 released

Version 1.3.5 is available for download (bug fix release).

Changes:

New nodes:
  • ROC view
  • Conditional Box Plot
  • Reference Row Filter
  • Reference Column Filter
Bug fixes:
  • Joiner: the first non-matching row got lost in right-outer and full-outer join
  • File Reader: data in Double and Int columns is now trimmed, accepting empty data as missing values there
  • File Reader: fixed reading from http-sources
  • Row ID: An error was reported if new column name existed in input table, even if no new column was appended
  • Row Sampling & Partition: the sampling and partitioning nodes now check for the class column if stratified sampling has been selected
  • Row Sampling: use seed if provided, was ignored before
  • Row Filter & String Replacer: are now able to accept "\n"
  • Row Filter: "." in regular expression now also matches newlines
  • Sorter: fixed dialog layout with long column names
  • Group-By: preserve original column spec if possible
  • Concatenate: fixed hilite propagation through the node
  • Histogram & Bar Chart: bar count fix
  • Histogram binner: rounding error fixed
  • Polynomial Regression: excluded columns containing missing values do not lead to an execution error any more
  • Database Writer: NULL values are now inserted into the database table for missing columns in input table
  • R: entered R code got lost if no R executable has been found
  • Molecule Sketcher: fixed a problem with drawing structures under Windows
  • Node descriptions: Fixed error message if fallback browser had to be used
  • BatchExecutor: temporary directories are now deleted under Windows if workflow was provided in a ZIP file
Enhancements:
  • Displaying of hydrogens in structure diagrams can be switched on/off in the CDK preference page
  • JavaScripting and Mathematical Expression nodes now compile the code only if it has changed and not upon every configure
  • More aggregation functions in Group-By node
  • BatchExecutor now allows input of encryption master key and commas inside node option values

15 January 2008: KNIME 1.3.3 released

Version 1.3.3 is available for download (bug fix release).

Changes:

New nodes:
  • Column Combiner
Bug fixes:
  • Static help pages, and tool to create help files based on node descriptions (no write access to installation directory required at runtime)
  • Reset confirm dialogs are enhanced by toggle "do not ask again"
  • Help content view: external links temporarily disabled (can crash application on Linux (eclipse bug))
  • Bit vector generator: equal bit set size for bit vectors parsed from id strings and no padding for bit vectors parsed from bit strings
  • File reader dialog: opening the dialog took very long if the last file read is on a network drive
  • SD reader: occasional null pointer exception
  • Smiles parser: input string is trimmed now
  • Weka classes generated GenericProperties.props file in user's home directory.
  • ThreadedColAppenderNodeModel shuffled columns if more than one column was replaced
  • OpenBabel node on windows: execution errors with more than ~5,000 molecules
  • added CTab type
  • Boxplot: bugfix and better progress report
  • String to number node: added dialog fields for thousands and decimal separator
  • Number to string node: integers are converted without decimal places now.
  • Sampling nodes: warning message if output table does not contain all classes because of too low row count (more classes than rows)
  • Sampling nodes: use of BitSet.size() instead of BitSet.cardinality() when counting output for stratified sampling
  • 2d coordinates were generated for all cdk columns (ignoring user settings)
  • Batch executor: Better help message on customizable java properties (old help resulted in error: "Unknown option '-Dorg.knime.core.maxThreads'")
  • Model Writer: appending automatically file extension ".pmml.gz" and print warning. Allow case-insensitive file extension (pmml and pmml.gz), like reader.
  • Core: fixed race condition in table ID generation code

09 January 2008: Feature article

Feature Article on "Workflow and pipeling tools" at QSAR World highlights KNIME.

21 November 2007: KNIME 1.3.1 released

Version 1.3.1 has finally hit the ground and is now available for download. This is a minor release with quite a number of new nodes, most of which are for data preprocessing but also include visualizer nodes and nodes for data mining. The list of new nodes includes:

  • case converter
  • cell splitter
  • column comparator
  • rewrite of joiner
  • low variance filter
  • many2one
  • one2many
  • number2string
  • string2number
  • string replacer
  • string replacer (dictionary)
  • group by
  • pivoting
  • numeric row splitter
  • pie chart
  • interactive pie chart
  • polynomial regression (learner & predictor)
  • linear correlation (computer & filter)
  • value counter
  • NaÏve Bayes (learner & predictor)
  • sota (self-organizing tree), learner & predictor
  • decision-tree learner (not based on weka)
  • multi-dimensional scaling
  • SVM (support vector machine, learner & predictor)
  • enrichment plotter
  • SDF properties export / import
  • CDK → Mol2/SDF/Smiles
  • R integration (allows local execution, more views)

Other improvements include:

  • all node descriptions integrated into Eclipse Help
  • hyperlinks activated in node description
  • tooltip at data outports shows table dimensions
  • bit vector cell (used for chemical fingerprint, e.g.) more feature rich (own renderer and icon)
  • great speedups in the hierarchical cluster node
  • leave-one-out functionality added to the cross validation node
  • full precision double renderer
  • tooltips in table view
  • more features in csv writer (separator char, decimal separator)
  • more features in file reader (short line support, file encodings)
  • more options in sampling node (stratified sampling)
  • more options in SDF/Mol2 reader (auto generate ID)
  • workflow editor: more intelligent node selection
  • workflow editor: pasting of nodes at current mouse position when triggered from popup
  • default dialogs allow for grouping of elements and horizontal layout
  • node description shown in text field when system fails to load web browser libraries (a common problem on some linux systems)
  • string2smiles node replaced by molecule type cast node

Changes in the developer version are, among others:

  • new node wizard allows user to create node in existing project (optional without sample code)
  • This only concerns developers that created their own dialog components by deriving from org.knime.core.node.defaultnodesettings.DialogComponent. One abstract method in this class was renamed (typo fix in the name). Thus, any derived class will not compile anymore. To fix the compile error, please rename the method "validateStettingsBeforeSave()" to "validateSettingsBeforeSave()".
  • More options to create output table, e.g.: exec.createConcatenateTable, exec.createJoinedTable
  • more options in ColumnRearrangeTable, allows swapping of columns

07 September 2007: External Tool and XLS Writer Plugin for KNIME 1.2.1 released.

Two additional nodes for KNIME 1.2.1 have been released. The External Tool Node enables the user to call an external executable from within KNIME, and the XLS Writer writes out data into a spreadsheet file. The two new plugins are available at the download page.

10 August 2007: KNIME Live! & Life Sciences!

Two KNIME workshops will take place in Konstanz, Germany, in November this year. See the workshop pages for details.

09 August 2007: Add-on package for KNIME 1.2.1 released

Two add-on plugins for KNIME 1.2.1 have been released. These plugins contain nodes which will be part of the upcoming version 1.3 but are already available as a preview working in v1.2.1.

14 May 2007: KNIME 1.2.1 released

KNIME 1.2.1 has been released and is available for download. This is a minor release with a couple of bug fixes and minor improvements:

Bug Fixes:

  • NPE in DataContainer when table only contains missing values (unreported)
  • Serialization of (Blob-) String > 64k failed (unreported)
  • inefficient duplicate checking in DataContainer (MAJOR CHANGE, bug #1090)
  • blob cells keeps reference to BLOB in SoftReference, fixes potential memory leak (unreported)
  • dead lock in filereader dialog (bug fix#1101)
  • "switch workspace" did not work in product, (bug fix#1067)
  • bad layout of aggregator in x-val meta node (bug fix#1080)
  • Weka dialogs failed to open on GNOME window manager, linux only, fix: forcing native java Look and Feel (unreported)
  • renderer in TableView was sometimes not updated when table structure changed (unreported)
  • HiliteHandler generation problem in NodeModel when model inports are present (unreported)
  • layout of missing value handling node messed up when adding many columns (bug fix #1114)
  • R-nodes: problem when result from server was invalid (unreported)
  • potential memory leak in node settings object (unreported)
  • Weka J48 node: problem on how to determine categorical columns (bug#1059)
  • Weka J48 node: some dialog options were ignored (unreported)
  • Cache Node: did not show progress information (unreported)

Improvements:

  • File-Reader has short line support, no rowID uniquification, progress
  • made default dialog components reusable and easier to extend
  • new batch executor (using Eclipse's OSGI class loading mechanism)
  • rewrite batch executor, supports now data cell options + nested elements

Chemistry related changes:

  • Upgrade to CDK 1.0 (still considered beta, though)
  • MoSS fix: switch from moss-4.6a to moss.4.31 (fixes bug #1088)
  • MoSS dialog: cleanup (meaningful default values), bug fix#1057
  • Babel node: failed when processing more than 2000 structures
  • added multiline string renderer to base chemistry types, removed proprietary SDF/Mol2Renderer, (bug #1092)
  • added getFallBackRenderer to all base molecule types; enables custom renderer registration
  • CDK: added global CDKNodePlugin.lock() method to be thread-safe (customizable through new preference page, all in response to bug #1089)
  • CDK (2D Generator): fewer warning messages, bug fix#1056
  • CDK: exporting all packages defined in this plugin (available in downstream plugins)
  • CDK (almost all nodes): many nodes had no dialog element in which the user could specify the name of the appended column (bug fix#1051)
  • CDK (Smiles Parsing): the smiles parser thread pool may have blocked the entire timeout if the smiles is easy (and quick) to parse (potential bug)
  • CDK: enabled Mol2->CDK translator node.
  • Mol2 Reader had problems with more than one structure in the file (unreported bug)
  • CDK: smiles parser ran into deadlock (bug fix #1089)

Other changes:

  • changed some of the utility classes (such as the table view classes, the StatisticsTable class, and the Normalizer class) to be used for the (new) correlation node
  • R-nodes: cleanup node description + dialog
  • Shape Manager Node: cleanup dialog layout
  • Sorter node: cleanup dialog layout

7 February 2007: KNIME 1.2.0 released

KNIME 1.2.0 has been released and is available for download. This is a major new release with numerous improvements (both visible and under the hood), many new nodes and a series of bug fixes. Among others, v1.2.0 now features:

  • a completely new look&feel,
  • support of the Eclipse update mechanism to install additional features,
  • several new visualization nodes and a rewrite of the existing nodes,
  • more native data mining nodes in addition to the Weka integration.
  • integration of reporting nodes based on the Eclipse BIRT project,
  • and expanded support of the Chemistry Development Kit (CDK) library.
We have also added a public forum for discussion of general questions and technical issues such as new node developments.

27 October 2006: KNIME 1.1.2 released

KNIME 1.1.2 has been released and is available for download. This is a bug fix release with the following changes:

  • bug fix: Missing Value Handling had corrupted dialog
  • bug fix: Cross-Validation node lost fold information
  • bug fix: Event listeners lost some events
  • bug fix: Tables did not support strings longer than 64k
  • bug fix: Scatterplot was sometimes unable to load the internal data after restart
  • bug fix: NodeView's "Close" button printed error message
  • enhancement: data base reader and writer support more types

18 September 2006: Tripos plans to support KNIME

Tripos, Inc. a leading provider of drug discovery informatics products and chemistry research, announced plans to support KNIME. The official press release is available here.

17 August 2006: KNIME 1.1.0 released

KNIME 1.1.0 has been released. There are no major changes in functionality compared to version 1.0.0; however, a few internals have changed:

  • Builds upon Eclipse 3.2, the currently latest version of Eclipse.
  • Renamed package names from de.unikn.knime... to org.knime... to reflect new ownership of KNIME.
    (If you have started to develop new nodes based on KNIME 1.0.0, these classes won't compile anymore. Fortunately, a simple reorganizing of imports in these classes will fix the problem.)
  • Re-organized plugin projects to clearly separate functionality (e.g. core classes and contributing nodes).
  • Cleanup of the chemistry plugin.
  • A few new nodes, e.g. a node to recompute the domain information of columns.
  • Minor bug fixes, among others a fix of a class loading problem when there are duplicate class definitions.

Other news

Valid XHTML 1.0 Strict

Clicky Web Analytics