{ "object_id": "https://biocomputeobject.org/GLY_000404/v-2.8.1", "spec_version": "1.3.0", "extension_domain": [ { "extension_schema": "https://www.w3id.org/biocompute/extension_domain/1.2.0/dataset/dataset_extension.json", "dataset_extension": { "additional_license": { "data_license": "https://creativecommons.org/licenses/by/4.0/", "script_license": "https://www.gnu.org/licenses/gpl-3.0.en.html" }, "dataset_categories": [ { "category_value": "Homo sapiens", "category_name": "species" }, { "category_value": "Protein", "category_name": "molecule" }, { "category_value": "Pharos cross-references", "category_name": "tag" }, { "category_value": "csv", "category_name": "file_type" }, { "category_value": "Reviewed", "category_name": "status" }, { "category_value": "internal", "category_name": "scope" } ] } } ], "provenance_domain": { "license": "\"https://creativecommons.org/licenses/by/4.0/\"", "name": "Human UniProtKB Xref Pharos", "contributors": [ { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Jeet Vora", "email": "jeetvora@gwu.edu" }, { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Robel Kahsay", "email": "rykahsay@gwu.edu" }, { "affiliation": "University of New Mexico", "contribution": [ "contributedBy" ], "name": "Tudor I Oprea", "email": "TOprea@salud.unm.edu" }, { "affiliation": "NIH", "contribution": [ "contributedBy" ], "name": "Paul Barrett", "email": "paul.barrett@nih.gov" }, { "affiliation": "NIH/NIGRI", "contribution": [ "contributedBy" ], "name": "Ajay Pillai", "email": "ajay.pillai3@nih.gov" }, { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Karina Martinez", "email": "karinamartinez@gwu.edu" }, { "name": "Luke Johnson", "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "email": "luke.johnson@email.gwu.edu" }, { "name": "Urnisha Bhuiyan", "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "email": "sbhuiyan28@email.gwu.edu" }, { "name": "Cyrus Au-Yeung", "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "email": "cyruschauyeung@gwu.edu" }, { "name": "Kate Warner", "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "email": "k.warner1@gwu.edu" } ], "created": "2020-03-20T14:14:17.733990", "modified": "2024-05-17T11:23:59.337Z", "version": "v-2.8.1" }, "description_domain": { "keywords": [ "Pharos", "Illuminating the Druggable Genome", "drug-targeted protein families", "G-protein-coupled receptors (GPCRs)", "ion channels (ICs)", "kinases" ], "pipeline_steps": [ { "name": "make-protein-dataset.py", "step_number": 1, "version": "", "input_list": [], "output_list": [], "prerequisite": [], "description": "The script extracts the UniProtKB accessions and protein name from the Pharos API and maps to the GlyGen's UniProtKB canonical accessions (masterlist) to generate a csv dataset. The accessions that are not mapped are logged in the log file. " } ] }, "execution_domain": { "external_data_endpoints": [], "environment_variables": {}, "script_driver": "manual", "software_prerequisites": [ { "version": "2.7.5", "name": "Python", "uri": { "uri": "https://www.python.org/download/releases/2.7.5/" } } ], "script": [ { "uri": { "uri": "https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py", "filename": "make-protein-dataset.py" } } ] }, "error_domain": { "empirical_error": {}, "algorithmic_error": {} }, "parametric_domain": [ { "step": "", "param": "", "value": "" } ], "usability_domain": [ "The Human UniProtKB Xref Pharos dataset contains human [taxid:9606] UniProtKB canonical accessions cross-referenced to Pharos UniProtKB accessions. The data is from Pharos API based on TCRD version 5.4.0. If you use this dataset please provide proper attribution to Pharos and GlyGen", "Pharos is the user interface to the Knowledge Management Center (KMC) for the Illuminating the Druggable Genome (IDG) program funded by the National Institutes of Health.", "The goal of KMC is to develop a comprehensive, integrated knowledge-base for the Druggable Genome (DG) to illuminate the uncharacterized and/or poorly annotated portion of the DG, focusing on three of the most commonly drug-targeted protein families: G-protein-coupled receptors (GPCRs), ion channels (ICs) and kinases", "Nguyen, D.-T., Mathias, S. et al, Pharos: Collating Protein Information to Shed Light on the Druggable Genome , Nucl. Acids Res.i>, 2017, 45(D1), D995-D1002. DOI: 10.1093/nar/gkw1072", "https://pharos.nih.gov/about" ], "etag": "7af2e8f13ec0ba2b6a78e86322819efb", "io_domain": { "input_subdomain": [ { "uri": { "filename": "batch.10.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.10.json" } }, { "uri": { "filename": "batch.11.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.11.json" } }, { "uri": { "filename": "batch.12.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.12.json" } }, { "uri": { "filename": "batch.13.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.13.json" } }, { "uri": { "filename": "batch.14.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.14.json" } }, { "uri": { "filename": "batch.15.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.15.json" } }, { "uri": { "filename": "batch.16.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.16.json" } }, { "uri": { "filename": "batch.17.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.17.json" } }, { "uri": { "filename": "batch.18.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.18.json" } }, { "uri": { "filename": "batch.19.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.19.json" } }, { "uri": { "filename": "batch.1.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.1.json" } }, { "uri": { "filename": "batch.20.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.20.json" } }, { "uri": { "filename": "batch.21.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.21.json" } }, { "uri": { "filename": "batch.2.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.2.json" } }, { "uri": { "filename": "batch.3.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.3.json" } }, { "uri": { "filename": "batch.4.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.4.json" } }, { "uri": { "filename": "batch.5.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.5.json" } }, { "uri": { "filename": "batch.6.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.6.json" } }, { "uri": { "filename": "batch.7.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.7.json" } }, { "uri": { "filename": "batch.8.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.8.json" } }, { "uri": { "filename": "batch.9.json", "uri": "https:/data.glygen.org/ln2downloads/pharos/03_2020/batch.9.json" } } ], "output_subdomain": [ { "mediatype": "csv", "uri": { "uri": "https:/data.glygen.org/ln2releases/data/v-2.8.1/reviewed/human_protein_xref_pharos.csv", "filename": "human_protein_xref_pharos.csv" } }, { "mediatype": "csv", "uri": { "uri": "https:/data.glygen.org/ln2releases/data/v-2.8.1/reviewed/human_protein_xref_pharos.stat.csv", "filename": "human_protein_xref_pharos.stat.csv" } } ] } }