{
    "bcoid": "GLY_001072",
    "title": "Arabidopsis Protein Names (NCBI RefSeq)",
    "description": "The Arabidopsis Protein Names (NCBI RefSeq) dataset contains Arabidopsis thaliana [taxid:3702] protein ...",
    "filename_list": "arabidopsis_protein_proteinnames_refseq.csv",
    "filename": "arabidopsis_protein_proteinnames_refseq.csv",
    "filetype": "csv",
    "categories": {
        "species": [
            "Arabidopsis thaliana"
        ],
        "molecule": [
            "Protein"
        ],
        "tag": [
            "Protein Names"
        ],
        "file_type": [
            "csv"
        ],
        "status": [
            "Reviewed"
        ],
        "scope": [
            "internal"
        ]
    },
    "minitable": {
        "content": [
            [
                "F4J8B4-1...",
                "NP_001319799..."
            ],
            [
                "Q8GXW6-1...",
                "NP_001319795..."
            ]
        ],
        "headers": [
            "uniprotkb_ca...",
            "refseq_ac..."
        ],
        "colwidth": [
            "40%",
            "20%"
        ]
    },
    "sampledata": {
        "type": "table",
        "data": [
            [
                {
                    "type": "string",
                    "label": "uniprotkb_canonical_ac"
                },
                {
                    "type": "string",
                    "label": "refseq_ac"
                },
                {
                    "type": "string",
                    "label": "refseq_protein_name"
                }
            ],
            [
                "F4J8B4-1",
                "NP_001319799",
                "F-box/LRR protein"
            ],
            [
                "Q8GXW6-1",
                "NP_001319795",
                "F-box/RNI-like superfamily protein"
            ]
        ]
    },
    "readme": "Name\n   - Arabidopsis Protein Names (NCBI RefSeq)\n\nVersion\n   - v-2.10.1\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   - \n\nChecksum\n   - \n\nUsability Domain\n   - The Arabidopsis Protein Names (NCBI RefSeq) dataset contains Arabidopsis thaliana [taxid:3702] protein names (primary and other alternative/synonym protein names) from NCBI RefSeq database. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen.\n\nDescription Domain\n   Keywords\n   - protein names\n   - NCBI RefSeq\n\n   Pipeline Steps\n   - Step-1: The script parses the NCBI RefSeq gpff files to extract the protein names and then maps the NCBI RefSeq accesion 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: arabidopsis_protein_xref_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.10.1/reviewed/arabidopsis_protein_xref_refseq.csv\n   Output Subdomain\n      Name: arabidopsis_protein_proteinnames_refseq.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.10.1/reviewed/arabidopsis_protein_proteinnames_refseq.csv\n      Name: arabidopsis_protein_proteinnames_refseq.stat.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.10.1/reviewed/arabidopsis_protein_proteinnames_refseq.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\n     Description: \"\"\n      Unique Values: \"26472\"\n      Name: \"refseq_ac\"\n     Description: \"Accession in the NCBI RefSeq database\"\n      Unique Values: \"26472\"\n      Name: \"refseq_protein_name\"\n     Description: \"Name of the protein in NCBI RefSeq database\"\n      Unique Values: \"11053\"",
    "downloadurl": "/ln2data/releases/data/v-2.10.1/reviewed/arabidopsis_protein_proteinnames_refseq.csv"
}