Section: Software and Platforms

KDD in Systems Biology

IntelliGO online

The IntelliGO measure computes semantic similarity between terms from a structured vocabulary (Gene Ontology: GO) and uses these values for computing functional similarity between genes annotated by sets of GO terms [83] . The IntelliGO measure is made available on line (http://plateforme-mbi.loria.fr/intelligo/ ) to be used by members of the community for exploitation and evaluation purposes. It is possible to compute the functional similarity between two genes, the intra-set similarity value in a given set of genes, and the inter-set similarity value for two given sets of genes.

WAFOBI : KNIME nodes for relational mining of biological data

KNIME (for “Konstanz Information Miner”) is an open-source visual programming environment for data integration, processing, and analysis. KNIME has been developed using rigorous software engineering practices and is used by professionals in both industry and academia. The KNIME environment includes a rich library of data manipulation tools (import, export) and several mining algorithms which operate on a single data matrix (decision trees, clustering, frequent itemsets, association rules...). The KNIME platform aims at facilitating the data mining experiment settings as many tests are required for tuning the mining algorithms. The evaluation of the mining results is also an important issue and its configuration is made easier.

A position of engineer (“Ingénieur Jeune Diplomé Inria”) was granted to the Orpailleur team to develop some extra KNIME nodes for relational data mining using the ALEPH program (http://www.comlab.ox.ac.uk/oucl/research/areas/machlearn/Aleph/aleph.pl ). The developed KNIME nodes include a data preparation node for defining a set of first-order predicates from a set of relation schemes and then a set of facts from the corresponding data tables (learning set). A specific node allows to configure and run the ALEPH program to build a set of rules. Subsequent nodes allow to test the first-order rules on a test set and to perform configurable cross validations. An Inria APP procedure is currently pending.

MOdel-driven Data Integration for Mining (MODIM)

Participants : Marie-Dominique Devignes [contact person] , Malika Smaïl-Tabbone.

The MODIM software (MOdel-driven Data Integration for Mining) is a user-friendly data integration tool which can be summarized along three functions: (i) building a data model taking into account mining requirements and existing resources; (ii) specifying a workflow for collecting data, leading to the specification of wrappers for populating a target database; (iii) defining views on the data model for identified mining scenarios. A steady-version of the software has been deposited through Inria APP procedure in December, 2010.

Although MODIM is domain independent, it was used so far for biological data integration in various internal research studies. A poster was presented at the last JOBIM conference (Paris, June 2011). Recently, MODIM was used by colleagues from the LIFL for organizing data about non ribosomal peptide syntheses. Feedback from users led to extensions of the software. The sources can be downloaded at https://gforge.inria.fr/projects/modim/ .