PhD position: Modeling biological knowledge with large language models

Within this project funded by the ADlab laboratory (Arnaud Droit Lab, compbio.ca), the main goal will be to transform the access and analysis of biological knowledge databases (BKDs) through the integration of these databases into a knowledge graph and the use of large language models (LLMs). This initiative aims to facilitate natural querying of biological data, making research more accessible to biologists without specific expertise in programming or data analysis. The main objectives include:

• Building a biological knowledge graph that integrates multiple BKDs to improve their cohesion and extract valuable information.
• Developing and training specific LLMs capable of interpreting and querying this graph in natural language.
• Creating and evaluating a conversational agent based on these LLMs, to allow intuitive querying of BKDs using natural language requests.

The selected candidate will be responsible for:

• Conducting in-depth research to effectively integrate BKDs into a coherent knowledge graph.
• Programming and training large-scale language models for the natural querying of biological data.
• Writing scientific papers to document the advancements and discoveries made during the project.
• Writing a doctoral thesis presenting significant contributions to the scientific community.
• Presenting research findings at national and international conferences, thereby contributing to the advancement of the field.

Ideal candidates for this position should possess:

• An MSc in Bioinformatics, Computer Science, Artificial Intelligence, or a related field.
• Strong programming skills, especially in Python and R, are essential for the development and training of models.
• Previous experience with databases, knowledge graphs, and/or large language models is highly recommended.
• A keen interest in molecular biology, bioinformatics, and biomedical research.
• Fluency in French and English (writing and speaking).