KI is part of the SciLake Cancer Knowledge Graph

The focus of this pilot project is to harness the power of SciLake services to aid in identifying biomarkers essential for personalised treatment and care, a crucial step towards the realisation of precision medicine. The case study of interest is Chronic Lymphocytic Leukemia (CLL), the most common adult leukaemia, characterised by a highly heterogeneous clinical course.

The goal of this pilot is to gain a better understanding of this heterogeneity by leveraging information from the CLL knowledge graph (KG), a cancer-specific KG that can effectively model the respective knowledge space using defined data sources. CLL KG will combine existing biomedical KGs like the Clinical Knowledge Graph with text mining and entity recognition services to reveal connections between clinical covariates, genes, proteins, metabolites, mutations, drugs, scientific literature, biomedical datasets, software and research consortia.

The potential applications of this cancer KG are numerous. On the one hand, it will enable researchers to query existing information to deepen their understanding of their findings. On the other hand, it will enable more advanced users to use the KG to uncover latent knowledge using more advanced graph mining algorithms and machine learning approaches. In the context of precision medicine, the CCL-KG will enable new discoveries of patient subtypes and why, for example, some groups respond better to certain treatments than others. In addition, researchers can use the KG as a benchmark to gauge the novelty of their discoveries in the field.

The teams have been diligently searching for existing knowledge graphs in the context of cancer. Recognising the ongoing need for an updated, cancer-specific knowledge graph, they have invested significant time in sourcing more general knowledge graphs from biology and precision medicine that will serve as a starting point. The challenge ahead lies in the effective integration and enrichment of these graphs to create a comprehensive cancer graph.

However, the potential benefits far outweigh the challenges. A successful cancer knowledge graph will be a significant asset for the wider cancer community. This tool, leveraging graph-based algorithms, user-friendly analytics, diverse clinical-biological databases, and drug-specific information, will empower scientists in their research.

Contacts

Leily Rabbani, Dania Machlab and Daniel Hägerstrand from the Clinical Genetics Group at the Department of Molecular Medicine and Surgery.