Interactome-Transcriptome Integration

Mgarcia et al (2010) companion website

M. Garcia, O. Stahl, P. Finetti, D. Birnbaum, F. Bertucci, and G. Bidaut. Linking interactome to disease: a network-based analysis of metastatic relapse in breast cancer. Handbook of Research on Computational and Systems Biology: Interdisciplinary Applications pp 406-427. L.A. Liu, D Wei, Y. Li, H. Lei Editors. IGI Global.

Breast Cancer Analysis (February 2010) [ITI-1.0]

The introduction of high-throughput gene expression profiling technology (DNA microarrays) in molecular biology and its expected application to the clinic has allowed the design of predictive signatures linked to a particular clinical condition or patient outcome in a given clinical setting. However, it has been shown that such signatures are prone to several problems:

  • (i) they are unstable and linked to the set of patients chosen as training.
  • (ii) data topology is problematic in regards to the data dimensionality (too many variables for too little samples).
  • (iii) diseases such as cancer are provoked by subtle misregulations which can’t be simply detected by current statistical methods..

    To find a predictive signature generalizable on multiple datasets, we devised a strategy of superimposition of a large scale protein-protein interaction data (human interactome) over several gene expression datasets (a total of 2464 breast cancers tumors were integrated), to find discriminative regions in the interactome (subnetworks) predicting metastatic relapse in breast cancer.

    This method, Interactome-Transcriptome Integration (ITI), was applied to several breast cancer DNA microarray datasets and allowed the extraction of a signature. Several combination for the training set were tested in runs 1-4 (all data, affy only, all but van de Vijver, all but Wang) and several signatures were generated.

    All subnetworks are available from the links listed below for exploration. The gene they contain have been linked to Gene Ontology and NCBI EntrezGene annotation databases for analysis. Exploration of annotations has shown this set of subnetwork reflects several biological processes linked to cancer and is a good candidate for establishing a network-based signature for metastatic relapse prediction in breast cancer.

    RUN-A1 - Without Van de Vijver (119 subnetworks)

    RUN-A2 - Without Wang (103 subnetworks)

    RUN-B1 - AFFYMETRIX(c) Without Van de Vijver (127 subnetworks)

    RUN-B2 - AFFYMETRIX(c) Without Wang (100 subnetworks)

    Other information

    Download source code and scripts here

    Questions? Comments? Do not hesitate to contact us from the Cibi home page.

    This page is Copyrighted (c) 2008-2013 Cibi.