On Molecular Biological Knowledge Graphs

Sameh K. Mohamed

Keywords: Knowledge Graphs, Molecular Biology, Link Prediction

TL;DR: On building and using knowledge graphs in the field of molecular biology

Abstract: Knowledge graph became a popular means for modelling data of interconnected entities i.e. linked data. They are adopted in many industrial and academic applications. In this work, we focus on the use of knowledge graphs in the field of molecular biology, where the linked data is centred around proteins and their associations with other biological entities. Recently, this type of knowledge graphs is becoming popular to support predicting different biological associations e.g. drug-protein targets, gene-disease associations, protein-protein interactions, etc. In this work, we explore the process of building and using these knowledge graphs. We first discuss available knowledge sources of molecular biology knowledge and the process of processing these sources to generate knowledge graphs. We then discuss various tasks and applications that can use these knowledge graphs to predict different biological associations. Finally, we provide an example for building and using knowledge graphs in molecular biology. We build a SwissProt-based knowledge graph of different protein associations, and we show by experiments that knowledge graph embedding models can achieve high accuracy in predicting the different protein associations compared to random baselines.

Contribution: Full research papers (8-12 pages)