TL;DR
The Monarch Initiative is a comprehensive, open-source platform that integrates vast amounts of genotype-phenotype data across multiple species. It addresses the challenge of connecting genetic variations to observable traits and diseases by leveraging sophisticated ontologies and providing powerful analytical tools, thereby accelerating mechanistic discovery and improving rare disease diagnostics.
What Problem Does This Paper Address?
Understanding the genetic basis of disease is a complex puzzle. Phenotypic outcomes in patients often don't neatly align with known diseases, candidate genetic variants might be in poorly understood genes, and research organisms may not perfectly mimic human conditions. Furthermore, environmental factors contributing to disease are often unknown or difficult to track. To connect the dots between an organism's genetic makeup, its observable characteristics (phenotypes), and diseases, researchers need to sift through numerous fragmented resources. The Monarch Initiative tackles this by creating a unified, integrative platform that bridges basic and clinical research, enabling sophisticated analysis and discovery across species.
Methods
The Monarch Initiative's core strength lies in its extensive use of ontologies to create a semantically rich knowledge graph. Key ontological resources developed or integrated include:
- Human Phenotype Ontology (HPO): A standardized vocabulary for clinical abnormalities, crucial for deep phenotyping in humans and enabling non-exact matching of phenotypic features to diseases.
- Unified Phenotype Ontology (uPheno): Integrates organism-specific phenotype ontologies using an Entity-Quality (EQ) approach, allowing for cross-species phenotype querying and the identification of animal models for human diseases. This involves a community-driven effort to reconcile and align these ontologies.
- Mondo: The unified disease ontology: A comprehensive ontology of diseases and disorders, integrating information from various sources like OMIM, Orphanet, and NCIt to provide a harmonized view.
- GENO (Genotype Ontology): Represents components of genotypes and their relationships, enabling integrated analysis of genotype-phenotype data across different resources and species.
- SEPIO (Scientific Evidence and Provenance Information Ontology): A framework for representing evidence and provenance behind scientific assertions, ensuring data quality and traceability.
The platform's architecture involves an ETL pipeline (Dipper) that ingests data from diverse sources, regularizes it according to the BioLink Model, augments it with ontology terms, and publishes it as RDF. This data is then integrated into a Neo4j graph database, which is queried and indexed by a Solr instance for fast access via the Monarch API.
Datasets
The Monarch Initiative integrates data from over 31 phenotype-related resources. These include:
- Research organism databases: MGD (Mouse Genome Informatics), ZFIN (Zebrafish Information Network), WormBase, FlyBase, IMPC (International Mouse Phenotyping Consortium), OMIA (Online Mendelian Inheritance in Animals), Animal QTLdb, Rat Genome Database (RGD), and Saccharomyces Genome Database (SGD).
- Human variant and gene-to-disease data: OMIM, ClinVar, Orphanet, and GWAS Catalog.
- Other genomic data types: Gene Ontology (GO) annotations, gene expression data (BgeeDB), protein-protein interaction data (BioGRID, STRING), pathway data (KEGG, Reactome), and chemical-disease associations (CTD).
- Cell line data: Coriell.
- Mouse data: Mouse Phenome Database and Mutant Mouse Resource and Research Centers (MMRRC).
The latest knowledge graph (September 2019) contains over 32.9 million nodes and 160 million edges, with a significant increase in gene-to-phenotype associations and disease models since the previous report.
Key Insights and Ideas
- Ontologies are the connective tissue: The paper strongly emphasizes that sophisticated, harmonized ontologies are not just descriptive tools but are fundamental for enabling powerful computational analysis, cross-species comparisons, and mechanistic discovery.
- Community-driven reconciliation is crucial: The success of uPheno highlights the importance of collaborative efforts to reconcile and align disparate phenotype ontologies, overcoming semantic incompatibilities.
- Bridging basic and clinical research: Monarch acts as a vital bridge, translating findings from model organisms into insights relevant for human disease diagnosis and research.
- The power of integrated knowledge graphs: By unifying diverse data sources into a single, queryable knowledge graph, Monarch unlocks new possibilities for hypothesis generation and data exploration that are impossible with fragmented resources.
- Standardization is key for translational impact: Monarch's involvement in initiatives like GA4GH and NCATS Translator underscores the need for standardized data models and APIs to facilitate data sharing and interoperability across the research and clinical landscape.
Key Results
- The Monarch platform integrates data from over 31 sources, covering more than 100 species.
- The September 2019 knowledge graph release contains over 32.9 million nodes and 160 million edges.
- There was a 44% increase in disease models since the 2017 report, reaching 26,433 models.
- The platform provides a powerful API that enables programmatic access to its integrated data and analytical capabilities.
- The new user interface (in beta) offers a simplified and intuitive way for researchers, clinicians, and the public to explore complex biological and disease data.
- Tools like Exomiser, powered by Monarch's data, are improving diagnostic rates for rare diseases.
Limitations and Future Work
- Data licensing complexity: The paper acknowledges that integrating data from multiple external sources, each with its own license and usage terms, presents complex implications for providing unified data access.
- Continued ontology development: The Monarch Initiative is continuously working on expanding and improving its ontologies, including integrating new species-specific phenotypes into uPheno and refining Mondo.
- Expanding data integration: Future work includes incorporating additional datasets, developing HCLS-compliant metadata, and extending the data model to include environmental exposures.
- Enhancing the user interface: The ongoing development of the new web interface aims to further improve user experience and accessibility.
- Cross-species scope: While Monarch focuses on human disease, there's a stated goal to expand its scope to include a higher diversity of species and domain expert contributions.
Summary updated