{
    "bcoid": "GLY_001270",
    "title": "Zebrafish Protein Citations (RefSeq)",
    "description": "The Zebrafish Protein Citations (RefSeq) datasets contains Danio rerio [taxid:7955] UniProtKB accessions ...",
    "filename_list": "zebrafish_protein_citations_refseq.csv",
    "filename": "zebrafish_protein_citations_refseq.csv",
    "filetype": "csv",
    "categories": {
        "species": [
            "Danio rerio"
        ],
        "molecule": [
            "Protein"
        ],
        "tag": [
            "Protein Citations"
        ],
        "file_type": [
            "csv"
        ],
        "status": [
            "Reviewed"
        ],
        "scope": [
            "internal"
        ],
        "file_category": [
            "citations"
        ]
    },
    "minitable": {
        "content": [
            [
                "Q6NZ04-1...",
                "The transcri..."
            ],
            [
                "F1QFD6-1...",
                "The coiled-c..."
            ]
        ],
        "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"
                }
            ],
            [
                "Q6NZ04-1",
                "The transcription factor six1 inhibits neuronal and promotes hair cell fate in the developing zebrafish (Danio rerio) inner ear.",
                "The Journal of neuroscience : the official journal of the Society for Neuroscience",
                "2006-10-11",
                "Bricaud Olivier, Collazo Andres",
                "protein_xref_pubmed",
                "17035528",
                "protein_xref_refseq",
                "NP_996978"
            ],
            [
                "F1QFD6-1",
                "The coiled-coil domain containing protein CCDC151 is required for the function of IFT-dependent motile cilia in animals.",
                "Human molecular genetics",
                "2014 Feb 1",
                "Jerber J, Baas D, Soulavie F, Chhin B, Cortier E, Vesque C, Thomas J, Durand B",
                "protein_xref_pubmed",
                "24067530",
                "protein_xref_refseq",
                "NP_001070837"
            ]
        ]
    },
    "readme": "Name\n   - Zebrafish Protein Citations (RefSeq)\n\nVersion\n   - v-2.12.1\n\nCreated on\n  - \n\nModified on\n   - \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 Zebrafish Protein Citations (RefSeq) datasets contains Danio rerio [taxid:7955] 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: zebrafish_protein_function_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/zebrafish_protein_function_refseq.csv\n   Output Subdomain\n      Name: zebrafish_protein_citations_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/zebrafish_protein_citations_refseq.csv\n      Name: zebrafish_protein_citations_refseq.stat.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.12.1/reviewed/zebrafish_protein_citations_refseq.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\"\n     Description: \"\"\n      Unique Values: \"1246\"",
    "downloadurl": "/ln2data/releases/data/v-2.12.1/reviewed/zebrafish_protein_citations_refseq.csv"
}