{
    "bcoid": "GLY_000036",
    "title": "Mouse Protein Citations (UniProtKB)",
    "description": "The Mouse Protein Citations dataset contains publication information for Mus musculus [taxid:10090] UniProtKB ...",
    "filename_list": "mouse_protein_citations_uniprotkb.csv",
    "filename": "mouse_protein_citations_uniprotkb.csv",
    "filetype": "csv",
    "categories": {
        "species": [
            "Mus musculus"
        ],
        "molecule": [
            "Protein"
        ],
        "tag": [
            "Protein Citations"
        ],
        "file_type": [
            "csv"
        ],
        "status": [
            "Reviewed"
        ],
        "scope": [
            "internal"
        ],
        "file_category": [
            "citations"
        ]
    },
    "minitable": {
        "content": [
            [
                "O35622-1...",
                "Diet1 functi..."
            ],
            [
                "O35622-1...",
                "Fibroblast g..."
            ]
        ],
        "headers": [
            "uniprotkb_ca...",
            "title..."
        ],
        "colwidth": [
            "40%",
            "20%"
        ]
    },
    "sampledata": {
        "type": "table",
        "data": [
            [
                {
                    "type": "string",
                    "label": "uniprotkb_canonical_ac"
                },
                {
                    "type": "string",
                    "label": "title"
                },
                {
                    "type": "string",
                    "label": "journal_name"
                },
                {
                    "type": "string",
                    "label": "publication_date"
                },
                {
                    "type": "string",
                    "label": "authors"
                },
                {
                    "type": "string",
                    "label": "xref_key"
                },
                {
                    "type": "string",
                    "label": "xref_id"
                },
                {
                    "type": "string",
                    "label": "src_xref_key"
                },
                {
                    "type": "string",
                    "label": "src_xref_id"
                }
            ],
            [
                "O35622-1",
                "Diet1 functions in the FGF15/19 enterohepatic signaling axis to modulate bile acid and lipid levels.",
                "Cell metabolism",
                "2013 Jun 4",
                "Vergnes L, Lee JM, Chin RG, Auwerx J, Reue K",
                "protein_xref_pubmed",
                "23747249",
                "protein_xref_uniprotkb",
                "O35622"
            ],
            [
                "O35622-1",
                "Fibroblast growth factor 15 functions as an enterohepatic signal to regulate bile acid homeostasis.",
                "Cell metabolism",
                "2005-10",
                "Inagaki Takeshi, Choi Mihwa, Moschetta Antonio, Peng Li, Cummins Carolyn L, McDonald Jeffrey G, Luo Guizhen, Jones Stacey A, Goodwin Bryan, Richardson James A, Gerard Robert D, Repa Joyce J, Mangelsdorf David J, Kliewer Steven A",
                "protein_xref_pubmed",
                "16213224",
                "protein_xref_uniprotkb",
                "O35622"
            ]
        ]
    },
    "readme": "Name\n   - Mouse Protein Citations (UniProtKB)\n\nVersion\n   - v-2.12.1\n\nCreated on\n  - 2019-01-23T17:21:05.903678\n\nModified on\n   - 2024-05-17T11:23:50.018Z\n\nCreated by\n   - Jeet Vora [jeetvora@gwu.edu], Robel Kahsay [rykahsay@gwu.edu], Karina Martinez [karinamartinez@gwu.edu], Luke Johnson [luke.johnson@email.gwu.edu], Urnisha Bhuiyan [sbhuiyan28@email.gwu.edu], Cyrus Au-Yeung [cyruschauyeung@gwu.edu], Kate Warner [k.warner1@gwu.edu]\n\nAuthors\n   - Maria Martin [martin@ebi.ac.uk], Preethi Vasudev [pvasudev@ebi.ac.uk]\n\nChecksum\n   - \n\nUsability Domain\n   - The Mouse Protein Citations dataset contains publication information for Mus musculus [taxid:10090] UniProtKB accessions that includes PMID, journal name, date and author and title. Publications of large scale studies, and high frequency publications pertaining to genome sequencing, cDNA, chromosomes have been filtered out. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen\n\nDescription Domain\n   Keywords\n   - PMID\n   - journal\n   - citations\n   - authors\n   - UniProtKB\n   - PubMed\n\n   Pipeline Steps\n   - Step-1: The script loads the UniProtKB publication data from the triplestore containing triples from nt file to retrieve the publication information associated with the UniProtKB accession and maps it to the UniProtKB canonical accessions. The script also filters out publications of large scale studies, and high frequency publications pertaining to DNA sequencing, cDNA, chromosomes using blacklisted PMIDs file.\n\nExecution Domain\n   Scripts   - https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-protein-dataset.py\n\n\n   Software Prerequisites\n      Name: Python\n      Version: 2.7.5\n      URI: https://www.python.org/download/releases/2.7.5/\n\nI/O Domain\n   Input Subdomain\n      Name: all_blacklisted_pmids.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/all_blacklisted_pmids.csv\n      Name: mouse_protein_blacklisted_pmids_uniprotkb.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/compiled/mouse_protein_blacklisted_pmids_uniprotkb.csv\n      Name: mouse_protein_function_uniprotkb.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_function_uniprotkb.csv\n      Name: mouse_protein_ptm_annotation_uniprotkb.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_ptm_annotation_uniprotkb.csv\n      Name: mouse_protein_site_annotation_uniprotkb.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_site_annotation_uniprotkb.csv\n      Name: uniprot-proteome-mus-musculus.nt\n      URI: https:/data.glygen.org/ln2downloads/ebi/2026_06_13/uniprot-proteome-mus-musculus.nt\n   Output Subdomain\n      Name: mouse_protein_citations_uniprotkb.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_citations_uniprotkb.csv\n      Name: mouse_protein_citations_uniprotkb.stat.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_citations_uniprotkb.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\"\n     Description: \"\"\n      Unique Values: \"19517\"",
    "downloadurl": "/ln2data/releases/data/v-2.12.1/reviewed/mouse_protein_citations_uniprotkb.csv"
}