Predictive Modeling of Plant Gene Expression via AI Analysis of Phase Separation and Chromatin Structure – PlantSCAPE

Meeting the nutritional needs of a global population that is expected to increase by 2.4 billion people over the next thirty years represents a major challenge for agriculture. This growth will require a significant increase in agricultural productivity, while taking into account the growing impacts of climate change. Heatwaves are severely affecting crop yields, particularly by interfering with flowering and fertilization—processes that are crucial for the production of seeds and fruits. Additionally, worsening droughts are reducing the viability of already marginal farmlands.

While past agricultural progress relied on varietal selection, mechanization, and irrigation, the current genomic era allows us to explore the genetic networks that govern agronomic traits. To adapt agriculture to climate challenges, it is essential to identify the genetic basis of stress tolerance and to understand how plants adapt to hostile environments.

As immobile organisms, plants have developed sophisticated mechanisms to perceive and adapt to environmental stress. Among these, chromatin-level regulation plays a key role in dynamically adjusting gene expression. Recent attention has focused on the three-dimensional architecture of the genome, which influences gene regulation in response to environmental and developmental signals. A newly emerging process at the heart of this organization is liquid-liquid phase separation (LLPS), which allows the formation of biomolecular condensates that concentrate proteins and nucleic acids, thereby modifying chromatin structure and finely regulating gene expression.

However, despite recent advances, fully leveraging this knowledge remains difficult, particularly due to the complexity of data generated by high-throughput sequencing. This is where artificial intelligence (AI) can play a transformative role by enabling the integration and interpretation of these complex datasets.

In this context, our project aims to harness the potential of AI to identify and characterize chromatin-associated proteins involved in the formation of biomolecular condensates, in order to precisely control gene expression in plants. To achieve this, we have established the PlantSCAPE-FK consortium (Plant chromatin Structure, Condensates, and AI-driven Prediction and Exploration), bringing together a French and a Korean research team.

With the support of AI, the project aims to build a predictive platform capable of identifying regulatory proteins across diverse plant genomes. Models will be used to detect the biophysical properties that promote phase separation. In parallel, gene expression models based on chromatin structure data will be developed to predict transcriptional activity according to genome architecture, including under stress conditions.

These discoveries will be followed by experimental validation of the identified regulators. In the long term, the project will explore the design of synthetic biology tools based on LLPS to finely modulate gene expression. By bridging artificial intelligence, chromatin biology, and genetic engineering, this project aims to lay the groundwork for an innovative framework for gene regulation in plants.