Research Day

(IB)2 Research Day 2018
16th November 2018 - 9.30am to 5.00pm

Groene Zaal, Congrescentrum U-Residence
VUB Campus
Generaal Jacqueslaan 271
1050 Brussels

9.30     9.35      Welcome

9.35     10.35    Keynote 1 - Carl Herrmann, University of Heidelberg

                        Multi-omics data integration for single-cell genomics
10.35   11.05    Charlotte Nachtegael, Unraveling the oligogenic potential

                         of developmental disorders

                         IB2, ULB

11.05   11.30    Coffee Break

11.30   12.00   Youssef Bouysran, Bioinformatic investigations of missense mutations

                         in the FBN1 gene

                         IB2, ULB

12.00   12.30    Nathaniel Mon Père, Quantitative Models for Cell Differentiation in Hematopoiesis

                         IB2, ULB


12.30   14.00    Lunch break and poster session


14.00   15.00    Keynote 2Hervé Isambert, Institut Curie Paris

                        Learning causal and non-causal networks from large scale genomic and

                        clinical data (Abstract below)

15.00   15.30    Gipsi Lima Mendez, Ocean Eco-system biology through integration

                         of heterogeneous data

                         IB2, ULB

15.30   16.00    Closing remarks / Future of the (IB)2

16.00   17.00    Musical Coffee

(IB)² - Interuniversity Institute of Bioinformatics in Brussels
Contact: Sophie de Buyl at, Matthieu Defrance matthieu.dc.defrance at



Learning causal and non-causal networks from large scale genomic and clinical data, Hervé Issambert, Curie Institute

Network reconstruction aims at disentangling direct from indirect dependences in information-rich data and has become ubiquitous to analyze the rapidly expanding resources of genomic and clinical data. However, most network inference methods are restricted to specific types of data and assume either causal or non-causal graphical models a priori. We have developed an information-based approach, which reconstructs causal, non-causal or mixed networks from large scale genomic or clinical data, without the need for an a priori choice on the causal or non-causal nature of reconstructed networks. Starting from a fully connected graph, it first removes dispensable edges by iteratively subtracting the most significant information contributions from indirect paths between each pair of nodes. The remaining edges are then filtered based on their confidence assessment or oriented based on the signature of causality in observational data. This computational approach outperforms or matches state-of-the-art methods for either causal (eg regulatory interaction) or non-causal (eg protein contact map) network reconstruction. In the talk, I will present different applications on a broad range of biological and clinical data, from single-cell transcriptomics and genomic alterations in tumor progression to long term evolution of vertebrates through whole genome duplication.