Research Project
10207002
Protein Knowledge Networks and Semantic Computing for Disease Discovery The growing volume and breadth of information from the scientific literature and biomedical databases pose challenges to the research community to exploit the content for discovery. This MIRA grant application will advance our knowledge mining and semantic computing system to accelerate data-driven discovery for understanding of gene-disease-drug relationships. We have employed natural language processing and machine learning approaches in a generalizable framework for bioentity and relation extraction from large-scale text. Our Protein Ontology supports protein-centric semantic integration of biomedical data for both human understanding and computational reasoning. We have also developed a resource to support functional interpretation and analysis of protein post-translational modifications (PTMs) across modification types and organisms. Building on our computational algorithms, bioinformatics infrastructure and community interactions, we will further develop literature mining tools to support automated information extraction across the bibliome and open linked data models for semantic integration of biomedical data from heterogeneous resources. Our text mining tools will be trained for different use cases using deep learning methods. We will develop RDF (Resource Description Framework) semantic models in an increasingly computable, inferable and explainable knowledge system to assist in hypothesis generation. We will present evidence in the form of textual artifacts and semantic models to ensure unbiased analysis and interpretation of results to promote rigorous and reproducible research. We will develop scientific case studies to drive the system development. Examples include PTM disease variant and enrichment analyses for drug target identification, genotype- phenotype knowledge mining for Alzheimer's Disease understanding, and gene-disease-drug knowledge network construction for COVID-19 drug repurposing. To foster community engagement, we will host workshops and hackathons to address critical fundamental research questions and emerging disease scenarios. We have fully adopted the FAIR (Findable, Accessible, Interoperable, Reusable) principles for resource sharing. All data, tools and research results will be broadly disseminated from the project website, accessible programmatically via RESTful API, queryable via SPARQL endpoints, and dockerized for community code reuse. The successful completion of this research will thus support scalable, integrative and collaborative knowledge discovery to accelerate disease understanding and drug target discovery.
