{ "io_domain": { "input_subdomain": [ { "uri": { "uri": "human_proteoform_glycosylation_sites_glyconnect.csv", "filename": "human_proteoform_glycosylation_sites_glyconnect.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_gptwiki.csv", "filename": "human_proteoform_glycosylation_sites_gptwiki.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_harvard.csv", "filename": "human_proteoform_glycosylation_sites_harvard.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_literature.csv", "filename": "human_proteoform_glycosylation_sites_literature.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_pdb.csv", "filename": "human_proteoform_glycosylation_sites_pdb.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_tyr_o_linked.csv", "filename": "human_proteoform_glycosylation_sites_tyr_o_linked.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "human_proteoform_glycosylation_sites_unicarbkb.csv" } }, { "uri": { "uri": "human_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "human_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "mouse_proteoform_glycosylation_sites_glyconnect.csv", "filename": "mouse_proteoform_glycosylation_sites_glyconnect.csv" } }, { "uri": { "uri": "mouse_proteoform_glycosylation_sites_pdb.csv", "filename": "mouse_proteoform_glycosylation_sites_pdb.csv" } }, { "uri": { "uri": "mouse_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "mouse_proteoform_glycosylation_sites_unicarbkb.csv" } }, { "uri": { "uri": "mouse_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "mouse_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "rat_proteoform_glycosylation_sites_glyconnect.csv", "filename": "rat_proteoform_glycosylation_sites_glyconnect.csv" } }, { "uri": { "uri": "rat_proteoform_glycosylation_sites_pdb.csv", "filename": "rat_proteoform_glycosylation_sites_pdb.csv" } }, { "uri": { "uri": "rat_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "rat_proteoform_glycosylation_sites_unicarbkb.csv" } }, { "uri": { "uri": "rat_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "rat_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "sarscov1_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "sarscov1_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "sarscov2_proteoform_citations_unicarbkb_literature.csv", "filename": "sarscov2_proteoform_citations_unicarbkb_literature.csv" } }, { "uri": { "uri": "sarscov2_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "sarscov2_proteoform_glycosylation_sites_unicarbkb.csv" } }, { "uri": { "uri": "sarscov2_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "sarscov2_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "hcv1a_proteoform_glycosylation_sites_literature.csv", "filename": "hcv1a_proteoform_glycosylation_sites_literature.csv" } }, { "uri": { "uri": "hcv1a_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "hcv1a_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "hcv1b_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "hcv1b_proteoform_glycosylation_sites_uniprotkb.csv" } }, { "uri": { "uri": "human_protein_xref_refseq.csv", "filename": "human_protein_xref_refseq.csv" } }, { "uri": { "uri": "mouse_protein_xref_refseq.csv", "filename": "mouse_protein_xref_refseq.csv" } }, { "uri": { "uri": "rat_protein_xref_refseq.csv", "filename": "rat_protein_xref_refseq.csv" } }, { "uri": { "uri": "sarscov1_protein_xref_refseq.csv", "filename": "sarscov1_protein_xref_refseq.csv" } } ], "output_subdomain": [ { "mediatype": "csv", "uri": { "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_pubchem_xref_mapping.csv", "filename": "protein_glygen_pubchem_xref_mapping.csv" } }, { "mediatype": "csv", "uri": { "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_pubchem_xref_mapping.stat.csv", "filename": "protein_glygen_pubchem_xref_mapping.stat.csv" } } ] }, "spec_version": "1.3.0", "object_id": "https://biocomputeobject.org/GLY_000499/1.5.30", "extension_domain": [ { "extension_schema": "http://www.w3id.org/biocompute/extension_domain/1.1.0/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": "", "category_name": "species" }, { "category_value": "Protein", "category_name": "molecule" }, { "category_value": "GlyGen-PubChem Protein Crossreferences", "category_name": "tags" }, { "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": "GlyGen-PubChem Protein Crossreferences Mapping", "contributors": [ { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Jeet Vora", "email": "jeetvora@gwu.edu" }, { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Rahi Navelkar", "email": "rsn13@gwu.edu" }, { "affiliation": "The George Washington University", "contribution": [ "createdBy" ], "name": "Robel Kahsay", "email": "rykahsay@gwu.edu" } ], "created": "2020-06-04T09:25:12.224023", "modified": "2020-06-04T13:29:28.212206", "version": "1.5.30" }, "description_domain": { "keywords": [ "PubChem Protein", "GlyGen", "crossreferences", "human", "mouse", "rat", "hcv1a", "hcv1b", "sarscov1", "sarscov2", "NCBI RefSeq" ], "pipeline_steps": [ { "name": "make-species-agnostic-dataset.py", "step_number": 1, "version": "", "input_list": [ { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_glyconnect.csv", "filename": "human_proteoform_glycosylation_sites_glyconnect.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_gptwiki.csv", "filename": "http://data.glygen.org/ln2wwwdata/reviewed/" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_harvard.csv", "filename": "human_proteoform_glycosylation_sites_harvard.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_literature.csv", "filename": "human_proteoform_glycosylation_sites_literature.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_pdb.csv", "filename": "human_proteoform_glycosylation_sites_pdb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_tyr_o_linked.csv", "filename": "human_proteoform_glycosylation_sites_tyr_o_linked.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "human_proteoform_glycosylation_sites_unicarbkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "human_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_glyconnect.csv", "filename": "mouse_proteoform_glycosylation_sites_glyconnect.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_pdb.csv", "filename": "mouse_proteoform_glycosylation_sites_pdb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "mouse_proteoform_glycosylation_sites_unicarbkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/mouse_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "mouse_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_glyconnect.csv", "filename": "rat_proteoform_glycosylation_sites_glyconnect.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_pdb.csv", "filename": "rat_proteoform_glycosylation_sites_pdb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "rat_proteoform_glycosylation_sites_unicarbkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "rat_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov1_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "sarscov1_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_citations_unicarbkb_literature.csv", "filename": "sarscov2_proteoform_citations_unicarbkb_literature.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_glycosylation_sites_unicarbkb.csv", "filename": "sarscov2_proteoform_glycosylation_sites_unicarbkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov2_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "sarscov2_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/hcv1a_proteoform_glycosylation_sites_literature.csv", "filename": "hcv1a_proteoform_glycosylation_sites_literature.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/hcv1a_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "hcv1a_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/hcv1b_proteoform_glycosylation_sites_uniprotkb.csv", "filename": "hcv1b_proteoform_glycosylation_sites_uniprotkb.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/human_protein_xref_refseq.csv", "filename": "human_protein_xref_refseq.csv" }, { "uri": "mouse_protein_xref_refseq.csv", "filename": "mouse_protein_xref_refseq.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/rat_protein_xref_refseq.csv", "filename": "rat_protein_xref_refseq.csv" }, { "uri": "http://data.glygen.org/ln2wwwdata/reviewed/sarscov1_protein_xref_refseq.csv", "filename": "sarscov1_protein_xref_refseq.csv" } ], "output_list": [ { "uri": "https://data.glygen.org/ln2wwwdata/reviewed/protein_glygen_uniprotkb_xref_mapping.tsv", "filename": "protein_glygen_uniprotkb_xref_mapping.tsv" } ], "prerequisite": [], "description": "The script loads the GlyGen's UniProtKB protein accessions for human, mouse, rat, hcv1a, hcv1b, sarscov1 and sarscov2 species and maps it to the refseq accessions and adds the corresponding glycosylation annotation to extract it into a csv dataset file" } ] }, "execution_domain": { "external_data_endpoints": [], "environment_variables": {}, "script_driver": "", "software_prerequisites": [], "script": [] }, "parametric_domain": [ { "step": "", "value": "", "param": "" } ], "usability_domain": [ "The GlyGen-PubChem Protein Crossreferences dataset contains UniProtKB glycoprotein accessions for human [taxid:9606], mouse [taxid:10090], rat [taxid:10116], hcv1a [taxid:11108], hcv1b [taxid:11116], sarscov1 [taxid:694009] and sarscov2 [taxid:2697049] glycoproteins for the purpose of PubChem Protein to GlyGen crossreferences. In the dataset, the UniProtKB glycoprotein accessions are mapped to the RefSeq accesions and glycosylation annotation corresponding to glycoprotein have been added. The glycosylation annotation provides information about total (N,O,S,C) glycosylation sites followed by (if available) reported N-glycans and/or reported O-glycans with the number of sites. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB, NCBI RefSeq and GlyGen" ], "etag": "8fe7b0efbbb6edd12413b5db16a37659", "_id": { "$oid": "5ed8f638f5c47b6196d34914" } }