This project deals with causal links detection in knowledge graphs representing multi-scale and multi-objectives problems, particularly present in life science. Several issues need to be addressed. First, life science ontologies involve many concepts and attributes, leading to a search space that is particularly large in such domains. Second, life science datasets may contain many imprecise and missing quantitative data, and uncertainty in scientific data and knowledge has to be accounted for. The third main concern relies on the interpretability of the results. A particular attention will be made in the project to help the end-users (i.e. the domain experts) to understand, evaluate and exploit the detected causal links.
These challenges will be applied in two distinct domains in life sciences that are related to environmental issues in plant development. Plant growth and development is tightly controlled by genotype, environmental cues and interaction between both (GxE). Phenotypes measured on the same genotype in different environments often demonstrate significant effects of the environment revealing a phenotypic plasticity. On the opposite, a robust phenotype can be considered as insensitive to the environment. In a context of climate change, phenotypic plasticity or robustness can confer adaptive values to organisms. That is a key challenging issue knowing the climatic warming preoccupation that has huge socio-economic, industrial and political impacts.
This research project aims to develop a new approach for automatic detection of gradual causal rules that express causality between different variables in knowledge graphs.
The developed approach will exploit time dependencies, contextual identity links and statistical methods. Targeted applications are thought to be decision-aid systems in the following domains: (i) the maize domain where the domain experts are interested in how the climatic cues differently impact the development of the different genotypes and therefore address the question of organisms’ adaptation to climate change; (ii) the rice domain where the domain experts are interested in determining the gene-gene interactions and their interactions with environments.