Topics and objectives :

Global population is rapidly increasing, representing a major challenge for food supply, exacerbated by climate change and environmental degradation. Most of those depend on agriculture, however plants health and survival are threatened by various biotic stressors. Despite the massive use of chemicals since the end of the Second World War, plant pathogens still represent a major cause of crop losses every year. While crop yield boost is needed, the food production system needs a strong revolution to massively reduce pesticide consumption leading to environmental pollution that is no longer viable. Advancing our knowledge about the functioning and success of plant-pathogens interactions is of prime importance to sustainably improve global plant health.

The biotechnological and digital advances of the last decade offer a great opportunity to overcome this stalemate. The flourishing of omics techniques has led to the possibility of studying complex biological systems, through systematic analysis of its content at the molecular level. A common paradox across scientific domains and particularly in biology, is the increasing ability to collect and create observational data far exceeding the ability to extract interpretable information and knowledge in this data deluge. Therefore, providing novel methodologies to address this challenge would represent a breakthrough improvement.

The objective of our research is to answer crucial open questions in the characterization of how plants respond to the biological environment (bio-aggressors and beneficial organisms) by developing multi-omics integration methods at the crossroad between network inference and machine learning.

Research axis :

Axis 1: to study plants response to biotic agent perturbations by constructing general frameworks to study unpaired, multi-modal, longitudinal and multi-omics data

Plant defense signaling mechanisms activated by pathogens is composed of several physiological, molecular, and metabolic changes mediated by a complex network of regulatory interactions that occur at different biological layers. Furthermore, plant pathogens have broad spectrum of characteristics spanning from ectoparasites to endoparasites, from biotrophic to necrotrophic, from avirulent to virulent strains, attacking different tissues from leaves to roots, varying time-life cycles, different genetics background, causing several different possible diseases, triggering different plant responses. Therefore, to globally characterize the modulation of plants molecular programme to respond to these wide-ranging attacks, it is necessary to perform multi-omics integration studies.

There are several challenges to accomplish this objective. Omics approaches generate large quantities of features and few or very few samples usually compose the object of the study, leading to the classical phenomenon in machine learning and bioinformatics named the curse of dimensionality. Furthermore, their correlation is usually very high, yielding redundant features. Also, irrelevant features that are not indicators for the phenotype under investigation, can be present and can mislead the learning algorithm and limit the model’s generalizability to unseen samples. Another issue is the class imbalance problem that induce bias in the model toward the majority classes and the danger of overfitting. Finally, another challenge to consider is the heterogeneity of each omics type due to the different high-throughput technologies used to collect the dataset. Therefore, we need to develop novel tailored methods to extract biological meaningful signatures from the complexity of these data.

Axis 2: To understand the modulation of the relationships among molecules upon interaction with biotic agents by inferring comprehensive plant-plant and plant-pathogens molecular networks

To investigate the relationship between molecules, network inference is used to study these complex interactions in a system biology framework. Indeed, biological network inference is a deeply studied problem in computational biology and several different approaches have been proposed. The connection among biomolecules (edges on the networks) can be inferred from the observational measurements and/or imputed by knowledge-driven approaches. Reconstructing interactions from observational data is a critical need for investigating natural biological networks, wherein network dimensionality (i.e. the number of interacting components) is usually high and interactions are time-varying. On the other side, knowledge driven networks can be limited by the paucity of known interaction among biomolecule especially for non-model plants. Since both methodologies have their advantages and disadvantages, we develop both approaches to tailor the methodology to the biological question.

Axis 3: To characterize plant molecular reprogramming on biological systems currently used at the Institut Sophia Agrobiotech & beyond

We are currently studying three phytopathosystems:

• Arachis wild species under biotic and/or abiotic stress. We focus on two wild relatives of peanuts plant Arachis stenosperma, highly resistant to the root-knot nematode Meloidogyne arenaria developing a hypersensitive response and moderately tolerant to drought stress and Arachis duranensis, that shows the opposite behaviour.  Wild species are known to be a good source of resistance genes, due to their high capacity to cope with different stresses.
• Arabidopsis thaliana in interaction with five distinct pathogens. A. thaliana is a model plant that is widely used in research and will allow a straightforward biological validation.
• Solanum lycopersicum is one of the most economically important vegetables throughout the world. Around 5 million hectares of tomatoes are grown worldwide annually producing more than 189.1 million metric tonnes (http://faostat.fao.org). During cultivation or in post-harvest storage, it is susceptible to more than 200 diseases caused by several different pathogens, such as viruses, viroids, fungi, oomycetes, bacteria, and nematodes that reduce yield and quality. Because of the large diversity of pathogens, the study of the tomato as a model for plant-pathogen systems will help to accelerate the discovery and understanding of the molecular mechanisms underlying disease resistance and offer the opportunity to improve the yield and quality of this crop.

Ongoing collaborations

• Institut Sophia Agrobiotech (Sophia-Antipolis, France): teams GAME, IPN, IPO, IRL.
• Inria (France): Dr Carriere, Data-shape team Centre Inria d’Université Côte d’Azur; Dr. Duvigneau, Acumes team Centre Inria d’Université Côte d’Azur; Dr. Sala, MUSCA team Inria and MaIAGE team INRAe, Université Paris-Saclay.
• University of Heidelberg (Germany): Dr Dugourd.
• EMPRAPA (Brasil): Dr Guimares.