{
    "bcoid": "GLY_000235",
    "title": "Mouse Protein Citations (RefSeq)",
    "description": "The Mouse Protein Citations (RefSeq) datasets contains Mus musculus [taxid:10090] UniProtKB accessions ...",
    "filename_list": "mouse_protein_citations_refseq.csv",
    "filename": "mouse_protein_citations_refseq.csv",
    "filetype": "csv",
    "categories": {
        "species": [
            "Mus musculus"
        ],
        "molecule": [
            "Protein"
        ],
        "tag": [
            "Protein Citations"
        ],
        "file_type": [
            "csv"
        ],
        "status": [
            "Reviewed"
        ],
        "scope": [
            "internal"
        ],
        "file_category": [
            "citations"
        ]
    },
    "minitable": {
        "content": [
            [
                "Q9D0E1-1...",
                "hnRNPM defic..."
            ],
            [
                "Q8CJ12-1...",
                "Novel epidid..."
            ]
        ],
        "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"
                }
            ],
            [
                "Q9D0E1-1",
                "hnRNPM deficiency leads to cognitive deficits via disrupting synaptic plasticity.",
                "Neuroscience letters",
                "2021 Apr 23",
                "Akinyemi AR, Li D, Zhang J, Liu Q",
                "protein_xref_pubmed",
                "33727124",
                "protein_xref_refseq",
                "NP_001103383"
            ],
            [
                "Q8CJ12-1",
                "Novel epididymis-specific mRNAs downregulated by HE6/Gpr64 receptor gene disruption.",
                "Molecular reproduction and development",
                "2007 May",
                "Davies B, Behnen M, Cappallo-Obermann H, Spiess AN, Theuring F, Kirchhoff C",
                "protein_xref_pubmed",
                "17034053",
                "protein_xref_refseq",
                "NP_001073317"
            ]
        ]
    },
    "readme": "Name\n   - Mouse Protein Citations (RefSeq)\n\nVersion\n   - v-2.12.1\n\nCreated on\n  - 2019-09-06T14:06:38.962804\n\nModified on\n   - 2024-05-17T11:23:53.104Z\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   - Not Applicable []\n\nChecksum\n   - \n\nUsability Domain\n   - The Mouse Protein Citations (RefSeq) datasets contains Mus musculus [taxid:10090] UniProtKB accessions mapped to publication information (PMID, title, journal name, publication, date, authors of corresponding RefSeq accessions).  If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen. \n\nDescription Domain\n   Keywords\n   - PMID\n   - journal\n   - citations\n   - authors\n   - RefSeq\n   - PubMed\n\n   Pipeline Steps\n   - Step-1: The script parses the NCBI RefSeq gpff files to extract the publication information. The script also maps RefSeq accessions to the UniProtKB Canonical accessions. \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_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_function_refseq.csv\n   Output Subdomain\n      Name: mouse_protein_citations_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_citations_refseq.csv\n      Name: mouse_protein_citations_refseq.stat.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/mouse_protein_citations_refseq.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\"\n     Description: \"\"\n      Unique Values: \"10856\"",
    "downloadurl": "/ln2data/releases/data/v-2.12.1/reviewed/mouse_protein_citations_refseq.csv"
}