Past Seminars

Here is the list of our past seminars :


Hervé Isambert (PCC, Institut Curie, Paris). Biophysics Seminar ENS-ESPCI - Hervé Isambert (PCC, Institut Curie, Paris)

Friday 17 November 2017 13:00-14:00 - Amphi Urbain, Ground Floor, Staircase N

Network reconstruction from multivariate information in genomic data

Network reconstruction has become ubiquitous to analyze the rapidly expanding, information-rich data of biological interest. However, most network learning methods are restricted to specific types of data and make an a priori choice on the causal or non-causal nature of the underlying model. We have developped an information-theoretic method which learns a large class of causal, non-causal or mixed networks from non-perturbative data, without the need for an a priori choice on the class of reconstructed network. Starting from a complete graph, the method iteratively removes dispensable edges, by uncovering significant information contributions from indirect paths. The remaining edges are then filtered based on their confidence assessment or oriented based on the signature of causality in observational data. The approach and associated algorithm, MIIC, outperforms or matches state-of-the-art methods for either causal or non-causal network reconstruction, such as gene regulatory networks or protein contact maps. Network reconstructions will be presented at different biological size and time scales, from single-cell gene expression data to gene and genome alterations in tumor development as well as long term evolution of vertebrates. MIIC is freely accessible at https://miic.curie.fr






Séminaires récents  (0)


Hervé Isambert (PCC, Institut Curie, Paris). Biophysics Seminar ENS-ESPCI - Hervé Isambert (PCC, Institut Curie, Paris)

Friday 17 November 2017 13:00-14:00 - Amphi Urbain, Ground Floor, Staircase N

Network reconstruction from multivariate information in genomic data

Network reconstruction has become ubiquitous to analyze the rapidly expanding, information-rich data of biological interest. However, most network learning methods are restricted to specific types of data and make an a priori choice on the causal or non-causal nature of the underlying model. We have developped an information-theoretic method which learns a large class of causal, non-causal or mixed networks from non-perturbative data, without the need for an a priori choice on the class of reconstructed network. Starting from a complete graph, the method iteratively removes dispensable edges, by uncovering significant information contributions from indirect paths. The remaining edges are then filtered based on their confidence assessment or oriented based on the signature of causality in observational data. The approach and associated algorithm, MIIC, outperforms or matches state-of-the-art methods for either causal or non-causal network reconstruction, such as gene regulatory networks or protein contact maps. Network reconstructions will be presented at different biological size and time scales, from single-cell gene expression data to gene and genome alterations in tumor development as well as long term evolution of vertebrates. MIIC is freely accessible at https://miic.curie.fr






Archives des anciens séminaires  (102)

Haut de page