Context

Nipah virus (NiV) is a recently emergent, highly pathogenic, zoonotic paramyxovirus first recognized following a 1998-99 outbreak of severe febrile encephalitis in Malaysia and Singapore (Chua et al., 2000). NiV can cause atypical pneumonia or necrotizing alveolitis with hemorrhage, pulmonary edema and aspiration pneumonia, leading to acute respiratory distress syndrome. As for the huge majority of risk group 4 pathogens, the knowledges on NiV virus infection remain very limited. Diagnosis, therapeutic and prophylactic means still do not exist. The Nipah project funded by the « Ministere de l’enseignement supérieur, de la recherche et de l’innovation » investigates these aspects in collaboration with Chinese institutions.

In this project, the SISTM team directed by Pr. Rodolphe Thiébaut aims at conducting the analysis and the modelling of the immune response to antiviral and vaccine strategies, using the data produced in pre-clinical and Phase I clinical, including immunological sub studies recording many biomarkers (cell phenotype, functionality, gene expression, antibody titers…).

SISTM is a team belonging to INSERM U1219 Bordeaux Population Health and INRIA Bordeaux Sud-Ouest research institutes. The group is dedicated to the analysis and the modelling of the data generated in epidemiology and medicine with a special focus on vaccines and immune interventions in HIV and other infectious diseases. Its expertise is mainly in biostatistics with a special emphasis on dynamical models based on ODE and statistical learning using moderately high dimensional data.

Job Description

As the SARS-CoV-2 crisis delayed almost all experiments for the Nipah project, the main objective of this postdoc position will focus on methods developments. Application of these methods to real datasets will also be possible thanks to Ebola projects (EBOVAC series) and Sars-CoV-2 projects (EMERGEN).

Model building is a crucial problem when modeling data using mechanistic models (see an example in Pasin et al. 2019). The mathematical model based on ordinary differential equations must be chosen and its identifiability must be verified. Then, a statistical model must be built on the parameters of the model to understand the link between 1/ the available descriptive variables and parameters 2/ the residual variability due to the heterogeneity of the observed individuals. Finally, the observation model allowing to link the data to the trajectories of the model need to be specified. Most of the model building strategy rely on the optimization of a penalized log-likelihood. We propose to build strategies around these topics including covariate model building, selecting the best penalization this covariate model building and down selecting parameters on which random effects are mandatory to model the inter-individual variability, Estimation will be based on likelihood optimization based on the SAEM algorithm as implemented in lixoft Monolix suite. All development will be made in R. Part of this work will be done in collaboration with Marc Lavielle from Inria Saclay Xpop at Ecole Polytechnique.

Other integrative analysis such as exploratory analysis may also be achieved on the data generated in the Nipah project. In particular, Principal component analysis (PCA) which is a technique for reducing the dimensionality of large datasets, increasing interpretability but at the same time minimizing information loss. Part of this work will be done in collaboration with Jérémie Guedj from Inserm IAME, Université de Paris.

The candidate will be integrated in a team of biostatisticians and modelers working on related topics: modeling of HIV vaccine response. The candidate will benefit from a very attractive environment with computing facilities and close collaborations with mathematicians (from INRIA and INSERM research centers) and immunologists (from the Labex Vaccine Research Institute).