{ "bcoid": "GLY_000396", "title": "Mouse Protein Submitted Names (UniProtKB)", "description": "The Mouse Protein Submitted Names (UniProtKB) dataset contains UniProtKB submitted names for mouse [taxid:10090] ...", "filename": "mouse_protein_submittednames.csv", "filetype": "csv", "categories": { "species": [ "Mus musculus" ], "molecule": [ "Protein" ], "tag": [ "Protein Submitted Names" ], "file_type": [ "csv" ], "status": [ "Reviewed" ], "scope": [ "internal" ] }, "minitable": { "content": [ [ "E9Q3P4-1...", "Centromere p..." ], [ "A0A087WRQ3-1...", "T cell recep..." ] ], "headers": [ "uniprotkb_ca...", "submitted_na..." ], "colwidth": [ "40%", "20%" ] }, "sampledata": { "type": "table", "data": [ [ { "type": "string", "label": "uniprotkb_canonical_ac" }, { "type": "string", "label": "submitted_name_full" }, { "type": "string", "label": "submitted_name_short" }, { "type": "string", "label": "ec_name" } ], [ "E9Q3P4-1", "Centromere protein F", "", "" ], [ "A0A087WRQ3-1", "T cell receptor alpha variable 1", "", "" ], [ "A5H0M4-1", "POTE ankyrin domain family, member G", "", "" ], [ "A5H0M4-1", "GSARP1", "", "" ], [ "S4R213-1", "Lactamase, beta-like 1", "", "" ], [ "A0A3Q4EGJ1-1", "Predicted gene 5155", "", "" ], [ "Q9D515-1", "Slx-like 1", "", "" ], [ "Q9D519-1", "Spermatogenesis associated multipass transmembrane protein 2b", "", "" ], [ "A0A0G2JEU8-1", "T cell receptor alpha variable 13N-1", "", "" ], [ "Q497P9-1", "Predicted gene 5941", "", "" ], [ "Q497P9-1", "Predicted gene, EG546347", "", "" ], [ "E9Q2C0-1", "Cilia and flagella associated protein 46", "", "" ], [ "L7N2D4-1", "Vomeronasal 2, receptor 29", "", "" ], [ "L7N2D0-1", "Predicted gene 3486", "", "" ], [ "L7N2D2-1", "Predicted gene 3278", "", "" ], [ "L7N2D9-1", "Predicted gene 3239", "", "" ], [ "L7N2D8-1", "Vomeronasal 2, receptor 71", "", "" ], [ "E9PZM7-1", "SR-related CTD-associated factor 11", "", "" ], [ "Q7TRP9-1", "Olfactory receptor Olfr654", "", "" ], [ "Q7TRP9-1", "Olfactory receptor family 52 subfamily U member 1", "", "" ], [ "Q7TRP7-1", "Olfactory receptor Olfr657", "", "" ], [ "Q7TRP7-1", "Olfactory receptor family 56 subfamily B member 1", "", "" ], [ "A0A0U1RQ84-1", "Zinc finger protein 626", "", "" ], [ "E9PUL3-1", "Chloride channel accessory 3B", "", "" ], [ "B1AY64-1", "Cysteine-rich perinuclear theca 3", "", "" ], [ "Q9D5T2-1", "RIKEN cDNA 4921524L21 gene", "", "" ], [ "D3Z649-1", "Potassium channel, subfamily T, member 2", "", "" ], [ "J3QP87-1", "Eukaryotic translation initiation factor 1A domain containing 14", "", "" ], [ "G3UY31-1", "Predicted gene, 21671", "", "" ], [ "J3QN64-1", "BTB domain containing 35, family member 22", "", "" ], [ "A0A2R8VJU7-1", "Septin 12", "", "" ], [ "E9Q7E4-1", "Predicted gene 8882", "", "" ], [ "Q9D583-1", "RIKEN cDNA 4930503E14 gene", "", "" ], [ "J3QPE5-1", "Predicted gene 5849", "", "" ], [ "Q80ZP0-1", "4930503B20Rik protein", "", "" ], [ "Q80ZP0-1", "RIKEN cDNA 4930503B20 gene", "", "" ], [ "F6UK53-1", "RIKEN cDNA 4933403O08 gene", "", "" ], [ "J3QNS5-1", "Major facilitator superfamily domain containing 4B4", "", "" ], [ "Q3UQ97-1", "Predicted gene 5460", "", "" ], [ "Q3UQ97-1", "Predicted gene, EG432838", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor MOR177-10", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor Olfr1141", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor family 5 subfamily W member 17, pseudogene 1", "", "" ], [ "A0A2I3BR05-1", "Predicted gene, 55960", "", "" ], [ "Q80W26-1", "cDNA sequence BC051070", "", "" ], [ "Q80W26-1", "Chitinase-like 6", "", "" ], [ "A0A1W2P738-1", "Predicted gene 6993", "", "" ], [ "A0A1W2P739-1", "Predicted gene 9639", "", "" ], [ "D3YUJ6-1", "Predicted gene 3183", "", "" ], [ "D3YUJ6-1", "Predicted gene, Gm16513", "", "" ], [ "D3YUJ3-1", "Cyclin Y-like 1", "", "" ], [ "A0A075B5V8-1", "Immunoglobulin heavy variable 1-47", "", "" ], [ "A0A075B5V7-1", "Immunoglobulin heavy variable V1-43", "", "" ], [ "A0A075B5V6-1", "Immunoglobulin heavy variable V1-42", "", "" ], [ "A0A075B5V5-1", "Immunoglobulin heavy variable 1-39", "", "" ], [ "A0A075B5V4-1", "Immunoglobulin heavy variable 1-37", "", "" ], [ "A0A075B5V3-1", "Immunoglobulin heavy variable 1-36", "", "" ], [ "A0A075B5V2-1", "Immunoglobulin heavy variable 1-34", "", "" ], [ "A0A075B5V1-1", "Immunoglobulin heavy variable 1-31", "", "" ], [ "A0A075B5V0-1", "Immunoglobulin heavy variable 1-26", "", "" ], [ "A0A075B5V0-1", "MCG114299", "", "" ], [ "Q3U4L8-1", "Zinc finger protein 51", "", "" ], [ "F7AG78-1", "Predicted gene 10959", "", "" ], [ "A0A1B0GT06-1", "Predicted gene, 29735", "", "" ], [ "A2A7B5-1", "PR domain containing 2, with ZNF domain", "", "" ], [ "Q7TQU7-1", "Olfactory receptor Olfr1357", "", "" ], [ "Q7TQU7-1", "Olfactory receptor 1357", "", "" ], [ "Q7TQU7-1", "Olfactory receptor family 1 subfamily I member 1", "", "" ], [ "A0A1B0GSF6-1", "Small integral membrane protein 32", "", "" ], [ "F7AEC7-1", "Predicted gene 2832", "", "" ], [ "Q6XAS2-1", "SSXB5 protein", "", "" ], [ "Q6XAS2-1", "Synovial sarcoma, X member B5", "", "" ], [ "Q8C2Y5-1", "F-box and WD-40 domain protein 14", "", "" ], [ "Q5SSJ1-1", "WAP four-disulfide core domain 17", "", "" ], [ "Q5SSJ1-1", "OTTMUSG00000000971 protein", "", "" ], [ "Q5SSJ1-1", "Activated macrophage/microglia WAP domain protein", "", "" ], [ "Q8VG58-1", "Olfactory receptor Olfr1230", "", "" ], [ "Q8VG58-1", "Olfactory receptor MOR237-1", "", "" ], [ "Q8VG58-1", "Olfactory receptor 1230", "", "" ], [ "Q8VG58-1", "Olfactory receptor family 4 subfamily C member 123", "", "" ], [ "A0A0G2JGK9-1", "T cell receptor alpha joining 37", "", "" ], [ "E9Q9H8-1", "Transmembrane 4 L six family member 19", "", "" ], [ "Q6S9I3-1", "HMW kininogen-II", "", "" ], [ "Q6S9I3-1", "Kininogen 2", "", "" ] ] }, "readme": "Name\n - Mouse Protein Submitted Names (UniProtKB)\n\nVersion\n - v-2.3.1\n\nCreated on\n - 2020-03-15T13:39:36.626348\n\nModified on\n - 2022-09-26T09:05:27.749986\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]\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 Submitted Names (UniProtKB) dataset contains UniProtKB submitted names for mouse [taxid:10090] UniProtKB/TrEMBL proteins accessions. In UniProt, UniProtKB/TrEMBL entries usually have 'Submitted name' which may later be improved by automatic annotation procedures but if not, it remains as provided by the submitter until the entry is manually annotated and integrated to UniProtKB/Swiss-Prot. The dataset is derived from 2019-09 UniProtKB release. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen\n\nDescription Domain\n Keywords\n - protein submitted name\n - protein name\n - UniProtKB\n\n Pipeline Steps\n - Step-1: The script loads the UniProtKB submitted names for protein accessions from the triplestore containing triples from nt file and extracts the recommended names into csv dataset file\n\nExecution Domain\n Scripts - https://github.com/glygener/glygen-backend-integration/blob/master/dataset-maker/extract-from-endpoint.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: uniprot-proteome-mus-musculus.nt\n URI: https:/data.glygen.org/ln2downloads/ebi/2023_10_30/uniprot-proteome-mus-musculus.nt\n Output Subdomain\n Name: mouse_protein_submittednames.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/mouse_protein_submittednames.csv\n Name: mouse_protein_submittednames.stat.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.3.1/reviewed/mouse_protein_submittednames.stat.csv\n\nFields\n Name: \"uniprotkb_canonical_ac\"\n Description: \"\"\n Unique Values: \"3397\"\n Name: \"submitted_name_full\"\n Description: \"The full submitted protein name in UniProtKB\"\n Unique Values: \"4454\"\n Name: \"submitted_name_short\"\n Description: \"The short submitted protein name in UniProtKB\"\n Unique Values: \"1\"\n Name: \"ec_name\"\n Description: \"Enzyme Commision name assigned to an enzyme by the Nomenclature Committee of the International Union of Biochemistry and Molecular Biology (IUBMB)\"\n Unique Values: \"3\"", "downloadurl": "/ln2data/releases/data/v-2.3.1/reviewed/mouse_protein_submittednames.csv" }