DIG-AI

Deciphering plant genotype-phenotype Interactions using knowledge Graphs and AI

Research Themes

This research focuses on understanding genotype-phenotype relationships in agronomic plants (rice, wheat, sorghum, Arabidopsis) using knowledge graphs and artificial intelligence. The objective is to structure heterogeneous biological data, enrich available knowledge, and identify candidate genes relevant to climate-change adaptation.

Main Scientific Axes

Axis 1. Data Integration and Knowledge Extraction

Develop methods for large-scale multi-source data integration in AgroLD (RDF), including dynamic integration strategies for high-volume datasets (xR2RML, SPARQL Micro-Services), and extract knowledge from scientific publications (NLP, NER, semantic relations).

Axis 2. Knowledge Enrichment in Graphs

Design strategies for knowledge-graph augmentation and entity alignment across heterogeneous sources, leveraging machine learning methods, embeddings, and recent LLM-based approaches.

Axis 3. Gene Function Prediction and Candidate Prioritization

Develop AI models to predict the function of poorly annotated or unknown genes, and rank genes most likely to be associated with target agronomic phenotypes, using knowledge graphs and deep-learning approaches.

Methodological Approach

  • AgroLD platform and Semantic Web standards (RDF, SPARQL, ontologies)
  • Integration of structured and unstructured data, including literature extraction
  • Learning models for entity alignment and graph enrichment
  • Deep-learning methods for function prediction and gene prioritization
  • Validation on real use cases with partners from the Global South

Expected Impact

This work aims to improve the use of multi-scale plant biology data, accelerate the identification of adaptation mechanisms, and provide operational tools for research-for- development in Global South contexts. Expected outputs include reusable knowledge resources, transferable AI methods, and biologically relevant hypotheses directly usable by geneticists and agronomists.

Application domains: rice, wheat, and tropical crops.