The last few years have seen major changes in experimental procedures, making it possible to gather huge quantities of data in diverse fields of biology. Scientists now routinely measure, characterize and localize an ever-growing number of molecules at the level of entire biological systems. Thus, despite the continuous expansion of “omic” approaches contributing to the elucidation of systems-level networks, we still know little about the organization of discrete biological activities in space and time, and their integration into larger systems and coherent phenotypes. The main difficulty lies in bridging the growing gap between high-throughput biological data production and analytical tools capable of developing a system level view of the data that also takes into account its biological context. Therefore, improving the assembly and interrogation of context-specific regulatory networks should increase our comprehension of the relationships environment-genotype–phenotype, with an increasing number of applications in the life sciences – from systems and synthetic biology to personalized medicine.
We propose to rationalize the comprehension of complex genotype–phenotype interactions through a “computational network biology” cycle, that we call “I3-BioNet”, composed of three building “blocks”: the inference (reconstruction), interrogation (characterization) and implementation (testing) of biological networks. The I3-BioNet Inference/Interrogation “blocks” are part of most systems and synthetic biology projects, but breakthroughs are still being made for each of these blocks. The Implementation module is only at its infancy. One of the under-appreciated applications of regulatory network reconstruction and interrogation is the formulation of new hypotheses and the generation of in-silico predictions which can be used to guide the design of new experiments, such as: i) Determining the functional effect of node(s) inhibition or activation in the inferred networks, e.g. transcription factors (TF); ii) Validation of TF complex formation and direct interactions between TFs and gene promoters; and iii) Metabolic and synthetic genome engineering, as envisaged in the new field of synthetic biology.
Many life science laboratories are acquiring such experimental protocols and automated experimental units. This trend will grow even stronger over the next decade. The iterative integration of experimentation and computation could lead to a virtuous circle. Over the last ten years, we have developed, adapted and gathered an arsenal of methods useful for this mission, ranging from machine learning, evolutionary computation, bioinformatics, and modeling, to visual analysis. I3-BioNet cycle will be modular, scalable and of generic value for tackling many questions in life sciences, from systems and synthetic biology to personalized medicine. We will focus strongly on major challenges facing society (related to health, industry and environment). The various types of wet-lab biology collaboration and data required are already available, and hypotheses/feedbacks have been generated. This should allow the systems biology workflow of the I3-BioNet project to be based on iterative cycles of quantitative data generation, model building, experimental testing and model refinement.
The cornerstone of I3-BioNet is the reconstruction of gene regulatory networks. Our team has been an active contributor to such methods, by combining results from data mining and computational biology with recent advances in statistical inference. More specifically, we will use and extend our tools i) to identify key regulatory elements of biological networks, including transcription factors, miRNA and their cis-regulatory binding sites (PreCisIon ) and the study of the interplay between genome organization and regulation ; ii) to reconstruct context-specific regulatory networks from expression data (LICORN , h-LICORN ); and iii) to make best use of the wide spectrum of -omics data that could be integrated to provide regulatory systems, including signaling and metabolism information to generate context-specific cell models (CoRegFlux ) .
 M. Elati, R. Nicolle, I. Junier, D. Fernandez, R. Fekih, J. Font, F. Képès. PreCisIon: PREdiction of CIS-regulatory elements improved by gene’s positION. Nucleic Acid Research, 41 (3): 1406-1415, 2013 PMID: 23241390
 C. Bouyioukos, M. Elati, F. Képès. Analysis tools for the interplay between genome layout and regulation. BMC Bioinformatics, 17 (5), 407, 2016 PMID: 27294345
 M. Elati, P. Neuvial, M. Bolotin-Fukuhara, E. Barillot, F. Radvanyi and C. Rouveirol. LICORN: learning co-operative regulation networks from expression data. Bioinformatics, 23:2407-2414, 2007 PMID: 17720703
 I. Chebil, R. Nicolle, G. Santini, C. Rouveirol and M. Elati*. Hybrid method inference for the construction of cooperative regulatory network in human. IEEE transactions on nanobioscience, 13: 97-103, 2014. PMID: 24771593
 D. Trejo, S. Khan, P. Trébulle, M. Elati. Networks interrogation for genotype-phenotype study of the yeast diauxic shift. International conference on systems biology (ICSB), to appear, Barcelone, 2016
This task is devoted to the systematic analysis of the network to obtain maximal insight from biological networks, investigating specific types of biological issues, such as: quantifying regulator activity and identifying key regulators (CoRegNet [1,2]); network modules /pathways extraction (PEPPER [4,5]); revealing potential mechanisms of a biological process or disease [2,5]; revealing the functions of nodes in the network [6,7]; and studying network cross-state/species deregulation/conservation .
 R. Nicolle, F. Radvanyi, and M. Elati. CoRegNet: reconstruction and integrated analysis of co-regulatory networks, Bioinformatics, 31 (18): 3066-3068, 2015 PMID: 25979476
 R. Nicolle, M. Elati, F. Radvanyi. Network transformation of gene expression for feature extraction. Machine Learning and Applications (ICMLA’11), IEEE, p108-113, Boca Raton, Florida, USA 2012
 E. Birmelé, M. Elati, C. Rouveirol, Ch. Ambroise. Identification of functional modules based on transcriptional regulation structure. BMC Proceedings, S4:1753-6561,2008 PMID: 19091051
 Ch. Winterhalter, R. Nicolle, A. Louis, C. To, F. Radvanyi, and M. Elati. PEPPER: Cytoscape app for Protein complex Expansion using Protein-Protein intERaction networks, Bioinformatics, doi: 10.1093, 30 (23): 3419-3420, 2014. PMID: 25138169
 C. To and M. Elati. A parallel genetic programming for single class classification. Genetic and evolutionary computation (GECCO), ACM Companion, 1579-1586, 2013.
 H. Njah, S. Jamoussi, W. Mahdi and M. Elati. A Bayesian approach to construct context-specific gene ontology. Computational Intelligence in Bioinformatics and Computational Biology (CIBCB), IEEE, to appear, 2016
 Koutroumpas, K., Dam, J. V., Toedt, G., Lu, Q., Reeuwijk, J. V., Boldt, K., … Elati, M., Képès, F. A systems biology approach towards the prediction of ciliopathy mechanisms, Cilia: 4(Suppl 1), 2015
 T. Picchetti, J. Chiquet, M. Elati, P. Neuvial, R. Nicolle, E. Birmelé. A model for gene deregulation detection using expression data, BMC Systems Biology, 9 S6, 2015 PMID: 26679516
This task aims to kick-start the I3-BioNet engineering cycle, by making use of the knowledge generated by the inference/interrogation blocks: i) to introduce an experimental design component into the I3-BioNet cycle, making it possible to propose high-quality functional experiments for scientists or a machine acting like a robot scientist (AdaLab project: www.adalab-project.org); and ii) to introduce computer-aided tools for the design of metabolic-transcriptional sub-networks with targeted behaviour and their genomic implementation (GREAT). This objective requires the development, adaptation and combination of state-of-the-art active learning, experimental design and optimization algorithms.
 C. Bouyioukos, F. Bucchini, M. Elati, F. Képès. GREAT: a web portal for Genome Regulatory Architecture Tools. Nucleic acids research, 44 (W1): W77-W82, 2016 PMID: 27151196
I3-BioNet SSB applications
We collaborate closely with life science investigators with expertise in various experimental methods and biological fields for which our approaches can bring new insights (to evaluate our algorithms rigorously, validate their predictions, and test their relevance to living systems). These applications provide proof-of-concept for the efficacy of the I3-BioNet cycle regardless of the organism, scale or studied process.
- Microorganisms and industrial biotechnology
- The diauxic shift in Saccharomyces cerevisiae : Ross King lab (Manchester)
- Metabolic engineering of Yarrowia lipolytica: Jean-Marc Nicaud lab (Micalis, INRA)
- Cancer and complex human diseases
- Bladder Cancer: François Radvanyi lab (Institut Curie)
- Epithelial tissues: Jenny Southgate lab (Univ. of York)
- Brain Cancer: Monique Dontenwill lab (Strasbourg)
- Malaria and infectious diseases: Rachida Tahar lab (IRD)
- Plant development and adaptation: Christophe Perin lab (CIRAD, Montpellier)