sur ce site

Home page > Equipes_Fr_En_It > EPC Inria Xpop : population modelling for life sciences

EPC Inria Xpop : population modelling for life sciences

Joint Team with the Centre de Recherche Inria Saclay - Île-de-France,
See the webpage of the Inria team Xpop.

Members of the team

- Marc Lavielle, Directeur de Recherche Inria

Permanent Researchers:
- Julie Josse, Partial time Professor, Ecole Polytechnique
- Erwan Le Pennec, P rofessor, Ecole Polytechnique
- Eric Moulines, Professor, Ecole Polytechnique

PhD students:
- Nicola Brosse
- Belhal Karimi
- Geneviève Robin
- Marine Zulian

- Katia Evrat

Research lines

Markov Chain Monte Carlo algorithms
While these Monte Carlo algorithms have turned into standard tools over the past decade, they still face diffculties in handling less regular problems such as those involved in deriving inference for high-dimensional models. One of the main problems encountered when using MCMC in this challenging settings is that it is diffcult to design a Markov chain that effciently samples the state space of interest. The Metropolis-adjusted Langevin algorithm (MALA) is a Markov chain Monte Carlo (MCMC) method for obtaining random samples from a probability distribution for which direct sampling is diffcult. MALA and its extensions have demonstrated to represent very effcient alternative for sampling from high dimensional distributions. We therefore need to adapt these methods to general mixed effects models.

Massive and missing data
The ability to easily collect and gather a large amount of data from di-erent sources can be seen as an opportunity to better understand many processes. It has already led to breakthroughs in several application areas. However, due to the wide heterogeneity of measurements and objectives, these large databases often exhibit an extraordinary high number of missing values. Hence, in addition to scienti-c questions, such data also present some important methodological and technical challenges for data analyst.
Missing values occur for a variety of reasons: machines that fail, survey participants who do not answer certain questions, destroyed or lost data, dead animals, damaged plants, etc. Missing values are problematic since most statistical methods can not be applied directly on a incomplete data. Many progress have been made to properly handle missing values. However, there are still many challenges that need to be addressed in the future, that are crucial for the users.

Population pharmacometrics
Pharmacometrics involves the analysis and interpretation of data produced in pre-clinical and clinical trials. Studies in pre-clinical and clinical pharmacology, pharmacokinetics, pharmacodynamics, and toxicology typically involve collection of various types of experimental data in individuals, groups and populations. Appropriate methods of analysis of such data requires an understanding of the underlying science including: biostatistics, computational methods and pharmacokinetic/pharmacodynamic modeling.
Population pharmacokinetics studies the variability in drug exposure for clinically safe and e-ffective doses by focusing on identi-fication of patient characteristics which signi-ficantly a-ect or are highly correlated with this variability. Disease progress modeling uses mathematical models to describe, explain, investigate and predict the changes in disease status as a function of time. A disease progress model incorporates functions describing natural disease progression and drug action. Natural disease progression is the change in disease status solely attributed to the progression of the disease. Drug action re-flects the e-ffect of a drug on disease status.

Model evaluation
Diagnostic tools are recognized as an essential method for model assessment in the process of model building. Indeed, the modeler needs to confront -his- model with the experimental data before concluding that this model is able to reproduce the data and before using it for any purpose, such as prediction or simulation for instance. The objective of a diagnostic tool is twofold: -first we want to check if the assumptions made on the model are valid or not ; then, if some assumptions are rejected, we want to get some guidance on how to improve the model.
We propose to develop new approaches for diagnosing mixed e-ffects models in a general context and derive formal and unbiased statistical tests for testing separately each feature of the model.

Precision medicine and pharmacogenomics
Pharmacogenomics involves using an individual’s genome to determine whether or not a particular therapy, or dose of therapy, will be e-ffective. Indeed, people’s reaction to a given drug depends on their physiological state and environmental factors, but also to their individual genetic make-up. Precision medicine is an emerging approach for disease treatment and prevention that takes into account individual variability in genes, environment, and lifestyle for each person. While some advances in precision medicine have been made, the practice is not currently in use for most diseases.
We therefore aim to develop methods and algorithms for the validation and selection of mixed e-ects models adapted to the problems of genomic medicine.

Models for intracellular processes (joint project with the Lifeware Inria team)
Signi-ficant cell-to-cell heterogeneity is ubiquitously-observed in isogenic cell populations. Cells respond di-fferently to a same stimulation. For example, accounting for such heterogeneity is essential to quantitatively understand why some bacteria survive antibiotic treatments, some cancer cells escape drug-induced suicide, stem cell do not di-fferentiate, or some cells are not infected by pathogens.
The main ambition of this project is to propose a paradigm change in the quantitative modelling of cellular processes by shifting from mean-cell models to single-cell and population models. The main contribution of Xpop focuses on methodological developments for mixed-e-ffects model identification in the context of growing cell populations.

mlxR, a R package for mixed effects models
Xpop is developing the mlxR package, a R package for the simulation and visualization of complex models for longitudinal data. The models are encoded using the model coding language MLXtran, automatically converted into C++ codes, compiled on the -fly and linked to R using the Rcpp package. That allows one to implement very easily complex ODE-based models and complex statistical models, including mixed e-ffects models, for continuous, count, categorical, and time-to-event data.
Fast simulation of realistic clinical trials is made possible with mlxR. Clinical trial simulation is the center piece of quantitative model based drug development (MBDD). Indeed, from repeated simulations of a certain trial with di-fferent treatments and di-fferent numbers of virtual patients, the probability of achieving a given target value, may be estimated in silico.

CMAP UMR 7641 École Polytechnique CNRS, Route de Saclay, 91128 Palaiseau Cedex France, Tél: +33 1 69 33 46 23 Fax: +33 1 69 33 46 46