The future challenge in animal production will be to provide food to a growing human population by respecting a balance between quality products, consumer acceptance and safety, as well as animal welfare. In a perspective of safe and sustainable food systems, reducing the use of antibiotics in livestock is a major concern. In fact, antibiotic resistance is one of the major medical challenges of the 21st century. The transfer of genes conferring resistance through the environment and the food chain, the potential for development of resistant bacteria and the appearance of therapeutic failures in human medicine, notably due to zoonotic bacteria, constitute major health issues for livestock farming sectors. In the pig breeding industry, the weaning period is often accompanied by a decreased growth rate caused by disparate food intake and diarrhoea due to digestive disorders that might be associated with bacterial population disequilibrium (i.e. dysbiosis) and/or opportunistic intestinal infections. Alarmingly, during this transition period the prophylactic use of antibiotics is still very frequent in order to limit piglet morbidity and mortality. Thus, reducing the prophylactic use of antibiotics in weaning pigs is a main issue and there is a strong need for alternatives. In this context, we have built a public-private partnership that gathers INRA scientists and industries from economic sectors of both animal feeding and pig breeding. PigletBiota is a precompetitive project that will study the physiological and genetic bases of the piglet sensitivity at weaning, as a prerequisite to identify innovative actions to adapt animals and pig production systems to a reduction of antibiotic use. The global aim of the PIGLETBIOTA project is to develop research that will contribute to adapt pig production systems to a reduction of antibiotics. The project proposes an integrative biology approach to determine the main factors influencing the variability of the individual’s robustness at weaning. We will monitor piglets for health, immune, stress and zootechnical traits and will characterize the intestinal microbiota diversity and composition as well as the contribution of host’s genotypes. The experimental design will combine various environments, including experimental and commercial farms, and ages at weaning and all animals will be fed without antibiotics. Animals (n~1000) will be clinically surveyed, measured for various traits related to production, immunity and stress, and genotyped with high-density SNP chips. The genetic parameters of the sensitivity at weaning will be estimated and genetic association studies performed. Faecal samples before and after the weaning date will be collected for characterizing the dynamics of the gut microbiota and studying its influence on the individual sensitivity at weaning. Animal and microbiota data will be vertically integrated in order to better understand the interplay between the these two levels of this biological system, and to develop robust indicators of weaning sensitivity. Finally, a functional screening using INRA platforms dedicated to human studies will be performed in order to detect active molecules to be tested in vivo and by using an axenic pigs model. The PigletBiota public-private consortium will favor translational research and innovation.

G protein-coupled receptors (GPCR) are very good targets for drugs. Their presence in the cell membrane make them accessible to drugs, and these receptors are involved in the vast majority of cellular processes. Indeed, GPCR are targeted by more than 30% of marketed drugs. To increase the efficacy and decrease adverse side effects of these drugs, a better comprehension of GPCR signalling is necessary. Knowledge concerning the different receptors has drastically increased these last years. The downside of this phenomenon is the profusion of omics data and scientific papers, the integration of which is a real challenge. The objective of ABLISS is the development of a method for building these signalling networks from available data as a whole: literature and large-scale datasets. The method will encompass two main components: (1) a natural language processing module, allowing to extract and format experimental results from scientific papers and (2) a knowledge-based method, allowing the inference of the network from these results. The framework will be applied to the deciphering of GPCR-triggered ß-arrestin- and ERK-dependent signalling. A first workpackage will be devoted to the knowledge-based method. The principle will be the formalization in ASP (Answer Set Programming) of the reasoning that allows the expert deducing network elements from experimental results. We have developed a first prototype, and thus demonstrated the feasibility of our approach. In ABLISS, we will extend the rules and predicate database to cover more experiment types, but also to adapt the reasoning module to the predicate-arguments structures that can be automatically generated by the natural language processing module. We will also study the reliability of a deduced fact. Finally, we will develop abductive reasoning to propose experimental plan allowing verifying hypotheses within the network. A second workpackage will concern the natural language processing module. During the preliminary work on knowledge-based network inference, the necessary manual extraction and formalization of experimental facts has appeared as a major limitation. We have shown, for a limited number of experimental results, that a transducer cascade allows extracting and formatting predicate-arguments structures directly from scientific publications. In ABLISS we will pursue this task, in particular through the development of a transducer cascade allowing extraction and formalization of experimental facts obtained through a large diversity of experiment types. Iteratively, we will ensure the completeness of this predicate ensemble. Finally, we will develop modules to complete the arguments of a predicate when these are not all present locally. The third workpackage will apply the framework to the building of ERK- and ß-arrestin-dependent signalling triggered by different GPCR. A first reason for this choice is that the concerned scientific publications corpus is relatively modest (around 1300 publications), allowing a manual control of obtained results. A second reason is the expertise of the partner coordinating the project in this particular area. New knowledge hypothesized in the network will be validated experimentally.

The worldwide emergence of antimicrobial resistant (AMR) bacteria relies on both the ability of mobile genetic elements (MGEs) to spread antibiotic resistance genes (ARGs), and the capacity of successful clones to disseminate. Evidence for the environmental origin of AMR in human and veterinary clinics highlights the mandate for the surveillance of emerging AMR. The objective of the PRE-EMPT project is to identify, and quantify the reservoir of mobile ARGs in different environments, and characterize the potential of these genes to be transferred to pathogens by combining high-throughput based techniques connecting the genes, the MGEs, and the bacterial communities. We will target three environmental sites, from urban, animal and littoral contexts, combine metagenomic techniques with enrichment methods (targeted PCR, hybridization capture, Hi-C), characterize the functional properties of the identified ARGs, and evaluate the dissemination of these ARGs in bacterial communities

Mutations in genes belonging to the RHO GTPase pathway are responsible for intellectual disability (ID), psychiatric disorders and brain development anomalies. The great heterogeneity of phenotypes associated with these gene mutations renders the development of therapeutic strategies strenuous. Studying PAK3, a central gene of the RHO GTPase pathway, will help us establish a genotype/phenotype correlation, which is essential to 1- define the rules behind mutation pathogenicity, 2- understand the underlying mechanisms and 3- propose adapted therapeutic approaches. 1- Our project is to define the genotype/phenotype correlation using about 20 different PAK3 mutations in order to understand the origin of PAK3-linked ID degree of severity, as well as why ID may sometimes be associated with other neurodevelopmental defects. Thus, we will establish and characterise the broadest cohort of patients bearing PAK3 mutations ever built. In parallel, we will assess the functional defects of mutated PAK3 variants and their effects on cell biology (shape, adhesion, migration) as well as neuron differentiation (neurite growth, dendritic spine formation). Our hypothesis states that mutation pathogenicity is not simply a loss or gain of function but may involve more complex mechanisms of signalling interference. Indeed, the presence of a mutated protein is often more deleterious than the lack of a protein. 2- To go further in analysing the severe forms of PAK3-linked ID, we created a new knock-in model bearing a mutation clinically responsible for a severe ID associated with secondary microcephaly. This mouse model presents strong behavioural and cognitive anomalies, as well as secondary microcephaly, reminiscent of the clinical case. Our project consists in a more thorough analysis of the mouse model behavioural and cognitive defects, in order to compare our results with the patient’s clinical traits. Our ex-vivo and in-vitro preliminary analyses allowed us to propose a new molecular mechanism of mutation pathogenicity, which we will investigate thoroughly. 3- We will test two phenotypical rescue strategies with the aim of further developing therapeutic solutions. The first strategy concerns severe forms of the disease. The degradation of stable pathogenic PAK3 proteins should, at least partially, restore phenotypic anomalies usually associated with severe ID. This strategy of specifically degrading stable pathogenic variants was never explored in the context of neurodevelopmental disorders, even while it is being developed as potential cancer treatment. It would also be applicable to over-activating mutations in genes belonging to the RHO GTPase pathway. The second rescue approach targets Cofilin, a convergence point of the RHO GTPase pathway. Several strategies targeting this actin polymerisation regulator were already explored to rescue behavioural anomalies and synaptic plasticity defects. We aim to demonstrate that this approach would also correct neuronal differentiation anomalies appearing during post-natal development. Thus, the efficiency of a cofilin-blocking peptide to restore neuritic arborisation and dendritic spine formation in mutated mice will be evaluated. This project is based on strong preliminary results and an already operational consortium composed of 2 clinician teams and 3 research teams (1 team being knowledgeable in the two fields). This project will allow us to understand the genotype/phenotype relations regarding PAK3 gene mutations as well as mutations on other genes belonging to the RHO-GTPase pathway. Our results will greatly help advance genetic counselling and patient monitoring. The post genomic and preclinical aspects of this project will also enable us to pave the way for new therapeutic approaches in the optic of personalised medicine.

In the last few years, neural models have allowed spectacular progress in natural language processing (NLP). The DeepTypo project proposes to use multilingual models of speech to design methods for automatically extracting, from audio recordings, typological information useful for language documentation and research (phonological and morphosyntactic complexity indices, similarities between languages…). Based on a collaboration between linguists and NLP researchers, the DeepTypo project sits squarely in the space of digital humanities by addressing fundamental questions of both communities. It will help linguists in their work of documenting and analyzing languages, especially “rare” or “poorly endowed” languages, by providing them with new tools and methods that will allow them, for example, to bring out new information on similarities between languages. Beyond the “tool development” aspect, the DeepTypo project aims, above all, at showing that the representations at the heart of neural networks can be used to answer fundamental questions in linguistic, by taking, as an example, current issues in creolistics (the study of creoles) and dialectology of Sino-Tibetan languages. Extracting typological information, the core of the DeepTypo project, will also contribute to the identification of the limits of fine-tuning. This approach has made it possible to develop, at low cost, NLP systems for several languages and many tasks and is often presented today as "THE" solution to all NLP problems. The identification of linguistic features captured by neural networks will allow us to verify if this is indeed the case: if a model is, for example, not able to detect and represent the tones of a language, it is more than likely that it cannot be used to develop a system for tonal languages. To achieve this ambitious goal, we will use neural representation analysis methods to interpret and understand the decisions of neural networks and will develop them along four original axes: 1. Based on the collaboration with the different partners of the project, we will try to identify richer features than those considered in the state of the art: if the existing works have focused on “simple” features (speaker gender, language of the utterance, ...), we will also consider information related to the diversity of the languages and to the linguistic characteristics of these languages (phonemic inventory, identification of tonal languages, ...). 2. In addition to existing analysis methods (e.g. linguistic probes), we will develop new methods to measure similarity between languages. Again, close collaboration between linguists and NLP researchers will be essential to define a linguistically relevant similarity (or similarities). 3. We will apply our methods to the 230 languages of the Pangloss collection (an archive of rare languages managed by LACITO) and to 15 creoles (collected mainly by LLL). These large-scale experiments will allow us to test state-of-the-art pre-trained models on languages with a wide variety of linguistic features rarely considered in NLP work. 4. We will apply these methods to language documentation support tasks, an application that has, until now, never been considered.