Seminars-Gene regulatory network inference from time series: machine learning approaches

EVENT : C3BI Seminars – Seminars-Gene regulatory network inference from time series: machine learning approaches


Speaker : Florence D’Alché-Buc, from Laboratoire de Statistiques et Applications, Télécom ParisTech, Paris Time : 2PM Date :   June 9th, 2016 Location : Room Jules Bordet  – METCHNIKOFF (67) ,Institut Pasteur, Paris


We consider the well known problem of gene regulatory network inference from time series measurements (gene expression, protein concentration) under the angle of Machine Learning. We address two different instances of this problem.

In the first instance we assume no information about the structure and no perturbed data: we develop a nonparametric dynamical model and a learning algorithm to identify the biological dynamical system and its network structure from the data. It gives us the opportunity to present a novel family of vector-valued models called operator-valued kernel based models. These models extends the well know family of scalar-valued kernel to multiple output prediction as well as structured output prediction.

If the results exhibit very good results compared to other methods, it is clear that the bottleneck of these methods is the lack of additional measurements especially, perturbation data that allow to identify the underlying system accurately. We prone the idea that when the goal of experiments is to estimate a model, data have to be chosen consequently. This introduces the second instance.

In this case, we address the design of experiments together with the identification of a parametric model (differential equations) of the network. The idea is to help the biologist to choose the experimental data, especially perturbation data like knock out, knock down, as well as the sampling rate of the measurements with a limited budget for experiments.The experimental design is thought as a one-player game and is solved using a active learning procedure that allows to both optimize the number and the cost of experiments to perform and the network parameters to be identified. On both problems, we show numerical results on the well known benchmarks of the DREAM Challenges. To conclude, new perspectives about dynamical modeling and machine learning and their use in computational biology are drawn.

N.B. The first work is a joint work with Néhémy Lim and Cédric Auliac, the second work is a joint work with Adel Mezine, Véronique Letort and Artemis Llamosi.


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Reconstructing a population’s past demography using whole genomes – 23 June 2016

EVENT : C3BI Seminars

Reconstructing a population’s past demography using whole genomes


Speaker : Flora Jay, from Laboratoire de Recherche en Informatique – Université Paris-Sud Time : 02:00 pm Starting Date : 23/06/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Reconstructing a population’s past demography using whole genomes

In population genetics, a key interest is to reconstruct the past history of a population using its genetic data. This history can be characterized by multiple demographic events such as migration of individuals, admixture with another population, or changes in population size (eg expansion or decline). With the availability of large-scale genomic data numerous methods have arose for untangling complicated demographic histories or retrieving a detailed picture of a population at different time periods.

Genomes are known to be extremely informative about demography and there are many ways to extract this information. I will focus on methods that infer past population sizes and mention a few significant recent developments. I will present our approach designed for an intermediate number of fully sequenced genomes. It relies on Approximate Bayesian Computation (ABC) which is a simulation-based statistical framework for generic model comparison and parameter inference. We demonstrate how the specificities of DNA sequencing data (namely haplotypic information, long range genetic correlation and genotyping errors) can be handle using ABC and fast genetic simulators, and further infer histories of successive bottleneck and expansions in human populations.


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Logical modelling of immune cell specification and reprogramming – 26 May 2016

EVENT : C3BI Seminars

 Logical modelling of immune cell specification and reprogramming


Speaker : Denis Thieffry, from Computational systems biology – Institut de Biologie de l’ENS Time : 02:00 pm Starting Date : 26/05/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Logical modelling of immune cell specification and reprogramming

Blood cells are derived from a common set of stem cells, which differentiate into more specific progenitors of erythroid, myeloid and lymphoid lineages, ultimately leading to functional cells. This ontogenesis is controlled by a complex regulatory network involving environmental signals, as well as transcriptional and epigenetic factors. Dynamical modeling of this network allows a better understanding of the control of cell fate, and to predict the effect of molecular perturbations. Using public data from molecular genetic experiments (qPCR, western blot, EMSA) or genome-wide essays (DNA-chip, ChIP-seq), we have built a comprehensive regulatory network encompassing transcription factors and signaling components involved in myeloid (macrophages and neutrophils) and lymphoid (B and T cells) development. Focusing on B cell and macrophage development, we have developed a model using a logical framework: the levels of activity of the different factors are assimilated to logical functions and their values are defined by a logical rule combining the levels of their regulators using logical operator (NOT, AND, OR). This framework enables a qualitative but nevertheless rigorous and predictive modeling of a regulatory network in the absence of quantitative data.

Our current model recapitulates several differentiation and experiments, including cytokines induced differentiation of common progenitors, pro-B cell reprogramming into macrophages induced by ectopic expression of transcription factors, and the effect of reported gene knock-downs. Tentatively, our model can be further use to design novel reprogramming experiments.


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Hierarchical Orthologous Groups – 12 May 2016

EVENT : C3BI Seminars

 Hierarchical Orthologous Groups: a unifying and scalable framework for large-scale gene evolution


Speaker : Christophe Dessimoz, from Laboratory of Computational Evolutionary Biology and Genomics – University of Lausanne

Time : 02:00 pm Starting Date : 12/05/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Hierarchical Orthologous Groups: a unifying and scalable framework for large-scale gene evolution

Sequencing data are rapidly piling up. Because many genes are highly conserved in sequence and function across different species—in some cases despite billions of years of intervening evolution—knowledge painstakingly gleaned through experiments can often be propagated across evolutionarily related genes. In theory, the more we know about the sequence universe, the easier elucidating these evolutionary relationships should get. Frustratingly however, the opposite seems true in practice: dealing with multiple species is conceptually and practically challenging, and as a result many evolutionary analyses remain stuck in a “two-species at a time” paradigm or only consider single-copy genes across multiple species. To overcome this impasse, I’ll introduce the concept of Hierarchical Orthologous Groups (HOGs)—nested groups of genes descending from a single ancestral gene within clades of interest. I’ll present an algorithm to accurately and efficiently infer HOGs. I will show how HOGs can be used to reconstruct ancestral genomes and to propagate functional knowledge from model species to non-model species.
Registrations for this seminars are now closed.

Seminars – From the exposome to the mechanome – 14 April 2016

EVENT : C3BI Seminars

From the exposome to the mechanome: Lessons learned and perspectives


Speaker : Marc Chadeau, from Faculty of Medicine – Imperial College London Time : 02:00 pm Starting Date : 14/04/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


From the exposome to the mechanome: Lessons learned and perspectives


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Tracking evolutionary changes with ancient DNA time capsules – 7 April 2016

EVENT : C3BI Seminars

Tracking evolutionary changes with ancient DNA time capsules


Speaker : Ludovic Orlando, from Natural History Museum of Denmark – University of Copenhagen Time : 02:00 pm Starting Date : 07/04/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Tracking evolutionary changes with ancient DNA time capsules

The survival of DNA molecules in long-dead fossil material offers a unique opportunity to catch evolution red-handed at the molecular level. This research area emerged in the mid-1980s and was for the first 25 years of its history, limited to the analysis of extremely limited genetic information. Recent technological breakthroughs have opened access to the complete genome sequence of ancient individuals and extinct species, such as the woolly mammoth and Neanderthals, and enhanced the time window for genome sequencing to at least one million years in permafrozen regions and half-a-million years in temperate caves. The information present in ancient genomes has considerably changed our understanding of the recent evolution of our own species, directly revealing patterns of population migration, admixture, selection and extinction. Additionally, the genetic characterization of ancient pathogens has unveiled the etiological agents of massive historical outbreaks, and the evolutionary arm race that progressively transformed such pathogens into massively deadly killers. Ancient DNA also helped track how humans became an evolutionary force, modifying their environment and transforming multiple animal and plant species into domesticates. As a diversity of past plant, bacterial, fungal and animal DNA material are still preserved in sediments and ice cores, ancient DNA can reveal how ancient communities and ecosystems were reshaped in the face of major environmental crises, thereby illuminating our understanding of ecological interactions and extinction processes. Ancient microbiomes, in particular the oral microbial communities, can also be reconstructed, providing a unique opportunity to follow the changes possibly introduced by recent cultural changes in our life-style. Beyond genomes, the profiling of epigenetic landscapes has become feasible and genome-wide nucleosome and methylation maps from past organisms have been reconstructed, paving the way for evaluating the evolutionary role of epigenetic reprogramming. I will present key developments in this recent revolution in ancient DNA research, using examples from my own research.


Registrations are closed

Seminars – Bioinformatics of structured RNAs beyond energy minimization – 24 March 2016

EVENT : C3BI Seminars

 Bioinformatics of structured RNAs beyond energy minimization


Speaker : Yann Ponty, from Computer Science Department (LIX) – École Polytechnique Time : 02:00 pm Starting Date : 24/03/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Abstract: For the last couple of decades, RNA Bioinformatics has been dedicating much effort to elucidate the sequence-structure relationship in structured non-coding RNAs at the secondary structure level. Quite noticeably, the field has experienced multiple paradigm shifts, transitioning from one-sequence-to-one-structure methods, exemplified by minimum free-energy structure prediction algorithms (MFold), to one-sequence-to-many-structures at the Boltzmann equilibrium (partition function), or postulating kinetic effects, and to many-sequences-to-one/many-structure(s), including comparative folding predicted from multiple homologous sequences. Yet, despite demonstrated improved performances, such methods only slowly permeate applied communities, partly due to their conceptual difficulty and to the absence of centralized tools and standard analysis protocols.

In this talk, I will briefly describe the three main paradigms in RNA Bioinformatics (energy minimization, thermodynamic equilibrium), and mention available state-of-the-art tools for structure prediction. I will mention the relative performances of such tools, and outline an analysis of the results using BRaliBase, the historic benchmark for (comparative) structure prediction. I will conclude with a short tale of benchmark design and interpretation.


Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – Human population genetics – 4 February 2016

EVENT : C3BI Seminars

 Human population genetics: genetic adaptation and epigenetic responses to environmental change


Speaker : Lluis Quintana-Murci, from Human Evolutionary Genetics Unit – Institut Pasteur Time : 02:00 pm Starting Date : 04/02/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Human population genetics: genetic adaptation and epigenetic responses to environmental change

Different environmental, demographic and selective forces, together with cultural and social characteristics of human lifestyle, shape the patterns of variability of the human genome at the population level. In particular, infectious diseases have been a major cause of human mortality, so natural selection is expected to act strongly on host defence genes. This is particularly expected for innate immunity genes, as they represent the first line of host defence against pathogens. I will present different cases of how some of these genes and the pathways they trigger have been targeted by natural selection, in its different forms and intensities, helping to delineate genes that are important for host defence, with respect to those exhibiting higher immunological redundancy. I will also discuss how population-specific genetic variation can profoundly impact immune-related molecular phenotypes, such mRNA and miRNA expression upon infection (expression quantitative trait loci – eQTL – mapping), and how these studies increase our understanding of immunological mechanisms under genetic control that have been crucial for our past and present survival against infection. Finally, I will discuss how the differences in lifestyle and habitat of human populations, together with their distinct patterns of genetic diversity, affect the epigenetic landscape of the human genome. Specifically, our studies of populations of African rainforest hunter-gatherers and sedentary farmers show that methylation variation associated with recent changes in habitat mostly involves immune functions, whereas that associated with historical lifestyle primarily affects developmental processes. Our work increases our understanding of whether and how populations are able to respond/adapt to environmental changes, including those related to pathogen pressures, over different time scales.

Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting

Seminars – P-Metagenomic Analysis Group – Paris – 29 January 2016

EVENT : C3BI Seminars – Large audience

P-MAG – Paris – Metagenomic Analysis Group


Speakers : Stevenn Volant research engineer, from Institut Pasteur, Amine Ghozlane researche engineer, from Institut Pasteur, Etienne Ruppé researcher from Hôpitaux Universitaire de Genève and Eric Pelletier researcher from Genoscope/CEA

    Time : 02:00 pm till 05:00 pm

      Starting Date : 29/01/2016

Location : Retrovirus room – LWOFF (22) ,Institut Pasteur, Paris


P-MAG_pasteur  

Registrations are now closed.

Seminars – Conservation and co-evolution – 21 January 2016

EVENT : C3BI Seminars

Conservation and co-evolution: from sequence analysis to protein-protein interactions


Speaker : Alessandra Carbone, from  Laboratory of Computational and Quantitative Biology, CNRS Université Pierre et Marie Curie, Paris

     Time : 02:00 pm     

Starting Date : 21/01/2016     

Location : Retrovirus room – LWOFF (22), Institut Pasteur, Paris


Conservation and co-evolution: from sequence analysis to protein-protein interactions

In computational biology, a fundamental question is the extraction of evolutionary information from DNA sequences. Here, we consider protein sequences and structures. Given a family of protein sequences and the associated distance tree, we shall explain how a fine reading of the conservation and co-evolution signals between residues in sequences can be used to identify protein binding sites, mechanical and allosteric properties, protein-protein interactions. Based on this novel approach to coevolution analysis, we reconstructed the protein-protein interaction network of the Hepatitis C Virus at the residue resolution. For the first time, coevolution analysis of an entire virus was realised, based on a limited set of protein sequences with high sequence identity within genotypes. The identified coevolving residues constitute highly relevant predictions of protein-protein interactions for further experimental identification of HCV protein complexes. The method can be used for interaction predictions for other viral protein interaction networks.

Due to security policy in Institut Pasteur, please register before if you plan to come to this meeting