{
    "bcoid": "GLY_000887",
    "title": "Dicty Proteoform Citations (GlyConnect)",
    "description": "The Dicty Proteoform Citations (GlyConnect) dataset provides citation details corresponding to the PubMed ...",
    "filename": "dicty_proteoform_citations_glycosylation_sites_glyconnect.csv",
    "filetype": "csv",
    "categories": {
        "species": [
            "Dictyostelium discoideum"
        ],
        "molecule": [
            "Proteoform"
        ],
        "tag": [
            "Phosphorylation Sites"
        ],
        "file_type": [
            "csv"
        ],
        "status": [
            "Reviewed"
        ],
        "scope": [
            "internal"
        ]
    },
    "minitable": {
        "content": [
            [
                "P04988-1...",
                "Occurrence o..."
            ],
            [
                "P04988-1...",
                "O beta-(N-ac..."
            ]
        ],
        "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"
                },
                {
                    "type": "string",
                    "label": "glytoucan_ac"
                }
            ],
            [
                "P04988-1",
                "Occurrence of N-acetylglucosamine-1-phosphate in proteinase I from Dictyostelium discoideum.",
                "The Journal of biological chemistry",
                "1980-08-10",
                "Gustafson G L, Milner L A",
                "protein_xref_pubmed",
                "6993483",
                "protein_xref_glyconnect",
                "118",
                "G68544GH"
            ],
            [
                "P04988-1",
                "O beta-(N-acetyl-alpha-glucosamine-1-phosphoryl)serine in proteinase I from Dictyostelium discoideum.",
                "Methods in enzymology",
                "1984",
                "Gustafson G L, Gander J E",
                "protein_xref_pubmed",
                "6438439",
                "protein_xref_glyconnect",
                "118",
                "G68544GH"
            ],
            [
                "P14328-1",
                "Multiple O-glycoforms on the spore coat protein SP96 in Dictyostelium discoideum. Fuc(alpha1-3)GlcNAc-alpha-1-P-Ser is the major modification.",
                "The Journal of biological chemistry",
                "2000 Apr 21",
                "Mreyen M, Champion A, Srinivasan S, Karuso P, Williams KL, Packer NH",
                "protein_xref_pubmed",
                "10766852",
                "protein_xref_glyconnect",
                "578",
                "G98043YN"
            ],
            [
                "P14328-1",
                "Multiple O-glycoforms on the spore coat protein SP96 in Dictyostelium discoideum. Fuc(alpha1-3)GlcNAc-alpha-1-P-Ser is the major modification.",
                "The Journal of biological chemistry",
                "2000 Apr 21",
                "Mreyen M, Champion A, Srinivasan S, Karuso P, Williams KL, Packer NH",
                "protein_xref_pubmed",
                "10766852",
                "protein_xref_glyconnect",
                "578",
                "G96881BQ"
            ],
            [
                "P14328-1",
                "Multiple O-glycoforms on the spore coat protein SP96 in Dictyostelium discoideum. Fuc(alpha1-3)GlcNAc-alpha-1-P-Ser is the major modification.",
                "The Journal of biological chemistry",
                "2000 Apr 21",
                "Mreyen M, Champion A, Srinivasan S, Karuso P, Williams KL, Packer NH",
                "protein_xref_pubmed",
                "10766852",
                "protein_xref_glyconnect",
                "578",
                "G68544GH"
            ],
            [
                "P52285-1",
                "Analysis of Skp1 glycosylation and nuclear enrichment in Dictyostelium.",
                "Glycobiology",
                "2001-04",
                "Sassi S, Sweetinburgh M, Erogul J, Zhang P, Teng-Umnuay P, West C M",
                "protein_xref_pubmed",
                "11358877",
                "protein_xref_glyconnect",
                "191",
                "G05897GU"
            ],
            [
                "P52285-1",
                "Analysis of Skp1 glycosylation and nuclear enrichment in Dictyostelium.",
                "Glycobiology",
                "2001-04",
                "Sassi S, Sweetinburgh M, Erogul J, Zhang P, Teng-Umnuay P, West C M",
                "protein_xref_pubmed",
                "11358877",
                "protein_xref_glyconnect",
                "191",
                "G23721DM"
            ]
        ]
    },
    "readme": "Name\n   - Dicty Proteoform Citations (GlyConnect)\n\nVersion\n   - v-2.4.1\n\nCreated on\n  - 2024-01-29T20:20:36\n\nModified on\n   - 2024-01-29T20:23:23\n\nCreated by\n   - Jeet Vora [jeetvora@gwu.edu], Robel Kahsay [rykahsay@gwu.edu], Karina Martinez [karinamartinez@gwu.edu], Jingyue Wu [jingyue.wu@email.gwu.edu], Luke Johnson [luke.johnson@email.gwu.edu], Urnisha Bhuiyan [sbhuiyan28@email.gwu.edu], Cyrus Au-Yeung [cyruschauyeung@gwu.edu], Jingyue Wu [jingyue.wu@email.gwu.edu]\n\nAuthors\n   - Frederique Lisacek [frederique.lisacek@sib.swiss], Julien Mariethoz [julien.mariethoz@sib.swiss]\n\nChecksum\n   - \n\nUsability Domain\n   - The Dicty Proteoform Citations (GlyConnect) dataset provides citation details corresponding to the PubMed ID (PMID). The script accesses the NCBI ENTREZ infrastructure to retrieve the publication information of the given PMID. The dataset is contributed by GlyGen and processed by GlyGen. If you use this dataset please provide proper attribution to GlyConnect (https://glyconnect.expasy.org/).\n\nDescription Domain\n   Keywords\n   - PMID\n   - journal\n   - citations\n   - authors\n   - GlyConnect\n   - PubMed\n\n   Pipeline Steps\n   - Step-1: The script parses the input file for PMIDs and accesses the NCBI Entrez infrastructure to download the publication information of the corresponding PMIDs\n\nExecution Domain\n   Scripts   - https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/make-proteoform-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: dicty_proteoform_glycosylation_sites_glyconnect.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/dicty_proteoform_glycosylation_sites_glyconnect.csv\n   Output Subdomain\n      Name: dicty_proteoform_citations_glycosylation_sites_glyconnect.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/dicty_proteoform_citations_glycosylation_sites_glyconnect.csv\n      Name: dicty_proteoform_citations_glycosylation_sites_glyconnect.stat.csv\n      URI: https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/dicty_proteoform_citations_glycosylation_sites_glyconnect.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\n     Description: \"\"\n      Unique Values: \"3\"\n      Name: \"title\"\n     Description: \"Title of the published article in the journal\"\n      Unique Values: \"4\"\n      Name: \"journal_name\"\n     Description: \"The name of the journal in which the article is published\"\n      Unique Values: \"3\"\n      Name: \"publication_date\"\n     Description: \"Contains the full date on which the issue of the journal was published. Sourced from MEDLINE/PubMed data element(fields) description.\"\n      Unique Values: \"4\"\n      Name: \"authors\"\n     Description: \"NA\"\n      Unique Values: \"4\"\n      Name: \"xref_key\"\n     Description: \"Source key associated to the  xref_id\"\n      Unique Values: \"1\"\n      Name: \"xref_id\"\n     Description: \"ID of the X-ref\"\n      Unique Values: \"4\"\n      Name: \"src_xref_key\"\n     Description: \"Source Xref  key\"\n      Unique Values: \"1\"\n      Name: \"src_xref_id\"\n     Description: \"Source X-ref ID\"\n      Unique Values: \"3\"\n      Name: \"glytoucan_ac\"\n     Description: \"Unique accession assigned to the registered glycan structure in GlyTouCan database\"\n      Unique Values: \"5\"",
    "downloadurl": "/ln2data/releases/data/v-2.4.1/reviewed/dicty_proteoform_citations_glycosylation_sites_glyconnect.csv"
}