Towards Discovery: An End-to-End System for Uncovering Novel Biomedical Relations

<p>This contains the embeddings as specified in the work: "<strong>Towards Discovery: An End-to-End System for Uncovering Novel Biomedical Relations"</strong>.</p> <p>The embeddings are created from 4 seperate knowledge bases:</p> <ul> <li><a href="https://ftp.expasy.org/databases/cellosaurus/" target="_blank" rel="noopener">Cellosaurus:&nbsp;</a>Concepts</li> <li><a href="https://ctdbase.org/downloads/" target="_blank" rel="noopener">CTD-Disease</a>: Concepts, Definitions, Synonyms</li> <li><a href="https://www.nlm.nih.gov/databases/download/mesh.html" target="_blank" rel="noopener">MeSH</a>: Concepts, Definitions, Synonyms. This includes embeddings for both the main and supplementary files, for Concepts and Definitions.&nbsp;</li> <li><a href="https://ftp.ncbi.nih.gov/gene/" target="_blank" rel="noopener">NCBI-Gene</a>: Concepts. Due ot the large size of the NCBI-cGene corpus we only provide the embeddings for the following top 7 organisms most prevalent in the corpus: <ul> <li>10090: house mouse&nbsp;</li> <li>10116: Norway Rat</li> <li>11676: human immunodeficiency virus (HIV)</li> <li>12814: Respiratory syncytial virus</li> <li>3702: thale cress</li> <li>7955: zebrafish</li> <li>9606: and human</li> </ul> </li> <li><a href="https://ftp.ncbi.nih.gov/pub/taxonomy/" target="_blank" rel="noopener">NCBI-Taxonomy</a>: Names</li> </ul> <p>The embeddings supplied were created using the <a target="_blank" rel="noopener">SapBERT</a> model, for the knowledge bases as of March 2024. The exact files used to generate the embeddings are provided as jsonl files.&nbsp;</p>