Biology is undergoing a historical revolution with the development of systems and synthetic approaches. Signaling pathways, transcriptional network and other cellular processes involve a large number of molecular actors with multiple interactions. While such processes are often well described mechanistically as lists of molecular reactions, not much is known on their quantitative and dynamical properties. Time-lapse fluorescent imaging of cells exposed to time-varying stimulus can be used to probe the dynamical behavior of gene circuits. Indeed, cells can be seen as complex machines, which response functions can be measured by time varying stimulations, as it is classically done in electrical and mechanical engineering. Several recent studies have used such methods to constrain the modeling of gene networks and signaling pathways. We are, however, far from being able to construct models of biological processes which are as predictive and robust as it is usually the case in physics and engineering. One of the main difficulties is the limited knowledge of the cell state that can be obtained simultaneously through fluorescence imaging and the existence of noise associated with gene transcription and translation. In turn, this strongly limits our ability to drive cellular processes, such as gene expression, over long time periods and with a quantitative accuracy. Recent developments in microfluidics, synthetic biology and optogenetics allow interrogating cellular processes in space and time at the single cell level. Having a mean to externally control, in real time, the expression level of a gene of interest, and use this to generate time-varying perturbations of the internal component of a regulatory network, would be a major step towards a quantitative understanding of how a cell functions. This would also have important consequences for applied biotechnology. As a matter of fact, a major challenge of synthetic biology is to engineer cells that can robustly perform a program in a broad range of environmental conditions and despite the stochastic nature of gene expression. However, given the complex, noisy nature of gene expression, an external control is usually needed to generate accurate time-varying perturbation of complex gene circuit for the interrogation of their behavior. The principle of controlling a dynamical system with a feedback-loop has been used extensively in engineering and is a key feature of most electromechanical devices of our everyday life. The basic idea is simple: monitor the readout and operate a change on the system to adjust it in real time so that it follows a given target profile. This permits to compensate for environmental fluctuations and un-modeled dynamics. We recently made a first step towards the construction of such a computer based control of gene expression in population of yeast cells. In this context, the CoGEx project aims at developing the experimental and theoretical tools for the computer-based remote-control of live cells and to use such a system to interrogate cellular processes at the single cell level. More specifically, our research project aims at (1) creating a versatile, open platform for the control of gene expression at the single cell level in yeast; (2) study the performance of real time control and (3) apply this method to find optimal conditions for the production of a biomolecule in an industrial fermentor. These research directions will be the basis for a larger, long term project that will aim at developing technologically and conceptually cell-machine interfaces based on genetics using advanced microfluidics, optogenetics, microscopy, control theory, modeling and synthetic biology.

The Web has become an essential part of our lives: billions are using Web applications on a daily basis and while doing so, are placing digital traces on millions of websites. Such traces allow advertising companies, as well as data brokers to continuously profit from collecting a vast amount of data associated to the users. At the same time, the users do not have any control of who is collecting their data and when. Even well-known privacy extensions fall short to protect user’s privacy entirely because companies use sophisticated techniques to track users on the Web. At the same time, website owners include a vast amount of third-party content and scripts in their websites, often to measure Web audience or for additional functionalities. However, website owners do not control whether such content is tracking their users. To give users more control over their data and hold website owners accountable for third-party trackers they include, the upcoming EU ePrivacy Regulation will make a significant transformation in the Web tracking ecosystem. ePrivacy will be based on the notion of user’s consent, which will impart users with an increasing control over their data. To technically express user’s consent in a Web browser, W3C has proposed a Do-Not- Track (DNT) standard. A remaining challenge is how the consent can be technically enforced in the existing Web applications without “breaking” them, while at the same time protecting the Web users. PrivaWeb aims at developing new methods for detection of advanced Web tracking technologies and new tools to integrate in existing Web applications that seamlessly protect privacy of users. In this project, we will integrate three key components into Web applications: privacy, compliance and usability. Our research will address methodological aspects (designing new detection methods and privacy protection mechanisms), practical aspects (large-scale measurement of Web applications, browser extensions implementation), and usability aspects (user surveys to evaluate privacy concerns and usability of existing and new protection tools).

The overall objective of PARDI is the formal, machine-supported verification of parameterized distributed systems. A parameterized system specification is a specification for a whole class of systems, parameterized by the number of entities and the properties of the interaction, such as the communication model (synchronous/asynchronous, order of delivery of message, application ordering) or the fault model (crash failure, message loss). To be able to verify such systems without explicitly instantiating all the parameters, new theoretical results and dedicated tools are required. Due to the fundamental undecidability of the problem, automatic tools are not powerful enough to verify rich parameterized systems: they are limited in their expressiveness and do not handle the variety of the interaction models. To assist and automate verification without parameter instantiation, PARDI uses two complementary approaches. First, a fully automatic model checker modulo theories is considered. Then, to go beyond the intrinsic limits of parameterized model checking, the project advocates a collaborative approach between proof assistant and model checker. This collaboration is two-ways: in one direction, the model checker is used as an explorer to generate elementary invariants that can be used and combined in interactive verification; in the other direction, a property which is needed by the proof assistant is discharged to the parameterized model checker. Cubicle, a model checker for array-based systems, and TLAPS, a proof assistant well-adapted to study distributed algorithms, are the basis of this project. Using case studies, both in parameterized distributed systems and in parameterized workflow-based systems, a theoretical analysis is required to exhibit the relevant parameters and to prove generic results (e.g. substitutability). This analysis is needed to define the richer data structures that will extend the model checker's expressiveness, and to inject these generic results into checkers and provers. A first contribution of the project will be the verification of parameterized distributed systems, for an arbitrary number of sites. A library of results for the many communication models and fault models which exist and their integration in the verification tools will be a second contribution. A third contribution will be the design and implementation of a verification toolchain based on a close interaction between a proof assistant and a model checker. A fourth contribution will be a new DSL for parameterized workflow-based systems with rich explicit parameterization features, and its translation into the checker and prover input language.

Neuroimaging and neuroscience, like many other experimental disciplines, currently face important challenges such as data management, the so-called replication crisis, and the reuse or sharing of data and analysis software tools. To address these challenges, the neuroimaging community started an international effort to standardize the sharing of magnetic resonance imaging (MRI) data, and more recently in 2018, data recorded with magnetoencephalography (MEG) and with electroencephalography (EEG) to be completed in 2019. This format, known as the Brain Imaging Data Structure (BIDS), now needs a wider adoption along with the development of dedicated software tools that operate seamlessly on BIDS formatted datasets. The project focuses on the less mature EEG and MEG BIDS ecosystem with three aims: 1) accelerate the research cycles by allowing analysis software tools to understand BIDS formated data, 2) simplify data sharing with high quality standards thanks to automated validation tools, 3) train French neuroscientists to leverage existing public BIDS MEG/EEG datasets and to share their own data with little efforts. To meet these aims the consortium proposes: - To consolidate the BIDS javascript validator for EEG/MEG. It allows simple validation of local files without any software installation besides a regular web browser. - To make the MNE-Python software used by dozens of labs in the world (approx. 500 citations in 4 years) more BIDS compatible. - To develop the first autonomous applications (a.k.a. BIDS apps) that can run a full MEG/EEG data analysis in a secured cloud. - To share datasets and state-of-the-art analysis pipelines from CEA Neurospin and ICM with the neuroscience community to establish standardized and reproducible data processing across labs (and accessorily promotes the French community) - To organize, disseminate tools and knowledge, via a satellite focusing on BIDS EEG/MEG software ecosystem during the CuttingEEG workshop happening in Marseille in 2020, as well as in other main neuroscience conferences worldwide (e.g. BIOMAG 2020, HBM, FENS). An engineer will be recruited for 18 months between Inria, CEA Neurospin and ICM.

To answer to the issues raised by the threats that may pose new technologies on privacy, lawyers and experts from many countries more and more emphasize on the principle of « privacy by design » (PbD). The basic principle of PbD is to take into account during the design phase of new products or systems, the need to protect privacy. From a technical point of view, the PbD has already been applied on an individual basis in specific areas. For the PbD can be adopted by industry, it is necessary to capitalize on these efforts, to identify the broad principles and to propose methods for their systematic application. The BioPriv project aims to propose such a systematic approach and apply it to a concrete case: the biometric access control. The followed approach is to design a formal model generic enough to take into account the relevant parameters for such systems (both in terms of technical requirements and from key players for the protection of privacy) and derive from this formal model procedures for analysis and generation of architectures satisfying the constraints of PbD. The expected results are of two types: methodological (systematic design method and evaluation) and applied (reference architectures). The results of the BioPriv project provide valuable assistance to architects who have to take into account the constraints of privacy protection as well as designers of future access control solutions. They will also facilitate, when appropriate, the implementation of certification procedures in relation to the requirements of private data protection.