Building a biological knowledge graph via Wikidata with a focus on the Human Cell Atlas

Abstract

The Human Cell Atlas is an international effort aiming at characterizing every cell type of the human body. Employing techniques such as single-cell RNA sequencing, mass cytometry, and multiplexed in situ hybridization, it will produce data from virtually all human tissues. This wealth of data can have a significant impact on biomedical research, but only if its content is genuinely available. Wikidata is a knowledge graph database emerging as a FAIR (Findable, Accessible, Interoperable and Reusable) repository for biological knowledge. The formatting and deployment of information from the Human Cell Atlas to Wikidata can increase information availability and impact, by inserting the findings in a network containing multiple associations of concepts of all areas of knowledge (within and outside Science). Conceptually defining cell types in a general and applicable concept, formalized into a database-compatible format, is a massive theoretical challenge. This PhD project aims at studying our current understanding of cell types for development a comprehensive ontological model in Wikidata for cell types. We will review the single-cell literature, refining and formalizing concepts for cell type delimitation. Furthermore, we will use Natural Language Processing and Machine Learning tools to automate knowledge extraction from scientific articles in the scope of the Human Cell Atlas. In an advanced step, we will apply concepts of network theory to develop tools for user-friendly querying of the database, making the knowledge ready for the academic community. (AU)