{ "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", "tags": "Protein Submitted Names", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "Q8VF63-1...", "Olfactory re..." ], [ "Q8VF63-1...", "Olfactory re..." ] ], "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" } ], [ "Q8VF63-1", "Olfactory receptor 881", "", "" ], [ "Q8VF63-1", "Olfactory receptor MOR162-7", "", "" ], [ "Q8VF64-1", "Olfactory receptor 883", "", "" ], [ "Q8VF64-1", "Olfactory receptor MOR162-6", "", "" ], [ "Q8VF64-1", "Olfactory receptor Olfr883", "", "" ], [ "E9Q3P4-1", "Centromere protein F", "", "" ], [ "A0A087WRQ3-1", "T cell receptor alpha variable 1", "", "" ], [ "A5H0M4-1", "POTE ankyrin domain family, member G", "", "" ], [ "A5H0M4-1", "GSARP1", "", "" ], [ "B1AQ89-1", "Predicted gene 11555", "", "" ], [ "B1AQ87-1", "Predicted gene 14190", "", "" ], [ "B1AQ85-1", "Keratin-associated protein 4-2", "", "" ], [ "A0A1Y7VJ58-1", "Predicted gene 7544", "", "" ], [ "A0A087WSM5-1", "Predicted gene 29049", "", "" ], [ "Q61406-1", "Complement factor H-related protein", "", "" ], [ "Q61406-1", "Complement factor H-related 1", "", "" ], [ "S4R213-1", "Lactamase, beta-like 1", "", "" ], [ "Q9D4U7-1", "RIKEN cDNA 4930556J24 gene", "", "" ], [ "Q9D4U4-1", "H2A histone family member L3", "", "" ], [ "A0A3Q4EGJ1-1", "Predicted gene 5155", "", "" ], [ "Q9D515-1", "Slx-like 1", "", "" ], [ "Q9D519-1", "Spermatogenesis-associated multipass transmembrane protein 2b", "", "" ], [ "D3YWW5-1", "Retrotransposon Gag-like 10", "", "" ], [ "A0A1Y7VK00-1", "Zinc finger protein 825", "", "" ], [ "Q497P9-1", "Predicted gene 5941", "", "" ], [ "Q497P9-1", "Predicted gene, EG546347", "", "" ], [ "B2RUK9-1", "Zinc finger protein 456", "", "" ], [ "J3QK46-1", "Transmembrane protein 200C", "", "" ], [ "A0A0G2JDF5-1", "T cell receptor alpha variable 14N-2", "", "" ], [ "A0A0G2JDF5-1", "T cell receptor alpha variable 14D-2", "", "" ], [ "A0A0G2JDF0-1", "T cell receptor alpha joining 31", "", "" ], [ "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", "", "" ], [ "G3X9Y6-1", "Aldo-keto reductase family 1, member C19", "", "" ], [ "Q7TRP7-1", "Olfactory receptor 657", "", "" ], [ "Q7TRP7-1", "Olfactory receptor Olfr657", "", "" ], [ "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", "", "" ], [ "E9Q913-1", "RIKEN cDNA G430095P16 gene", "", "" ], [ "E9Q912-1", "RAP1, GTP-GDP dissociation stimulator 1", "", "" ], [ "D3Z642-1", "RIKEN cDNA 4930595D18 gene", "", "" ], [ "D3Z649-1", "Potassium channel, subfamily T, member 2", "", "" ], [ "J3QP87-1", "Predicted pseudogene 2035", "", "" ], [ "E9PWM3-1", "Armadillo repeat-containing, X-linked 4", "", "" ], [ "J3QN64-1", "BTB domain-containing 35, family member 22", "", "" ], [ "Q50E62-1", "Aromatic-preferring amino acid transporter", "", "" ], [ "Q50E62-1", "Solute carrier family 7 (cationic amino acid transporter, y+ system), member 15", "", "" ], [ "E9PX79-1", "Membrane-associated ring-CH-type finger 10", "", "" ], [ "Q3UTQ6-1", "Zinc finger protein 868", "", "" ], [ "A0A2R8VJU7-1", "Septin 12", "", "" ], [ "E9QAJ8-1", "Predicted gene 3468", "", "" ], [ "E9Q7E4-1", "Predicted gene 8882", "", "" ], [ "F6WAU7-1", "Zinc finger protein 945", "", "" ], [ "E9Q6Y8-1", "Ubiquitin-specific peptidase 31", "", "" ], [ "F6XLV1-1", "Ciliary rootlet coiled-coil, rootletin family member 2", "", "" ], [ "E9Q6Y4-1", "Zinc finger protein 94", "", "" ], [ "A0A286YDB2-1", "RIKEN cDNA 4930469K13 gene", "", "" ], [ "Q8C8Y3-1", "RIKEN cDNA A930002H24 gene", "", "" ], [ "J3QPE5-1", "Predicted gene 5849", "", "" ], [ "Q80ZP0-1", "4930503B20Rik protein", "", "" ], [ "Q80ZP0-1", "RIKEN cDNA 4930503B20 gene", "", "" ], [ "J3QNS5-1", "Major facilitator superfamily domain-containing 4B4", "", "" ], [ "Q3UQ97-1", "Predicted gene 5460", "", "" ], [ "Q3UQ97-1", "Predicted gene, EG432838", "", "" ], [ "Q9D7P3-1", "Keratin-associated protein 4-13", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor MOR177-10", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor 1141", "", "" ], [ "Q8VFQ7-1", "Olfactory receptor Olfr1141", "", "" ], [ "A0A0U1RPM1-1", "Predicted gene, 36028", "", "" ], [ "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", "", "" ], [ "E9Q271-1", "WD repeat domain 17", "", "" ], [ "D3YUJ3-1", "Cyclin Y-like 1", "", "" ] ] }, "readme": "Name\n - Mouse Protein Submitted Names (UniProtKB)\n\nVersion\n - 1.8.24\n\nCreated on\n - 2020-03-15T13:39:36.626348\n\nModified on\n - 2020-03-16T11:15:40.779396\n\nCreated by\n - Jeet Vora [jeetvora@gwu.edu], Rahi Navelkar [rsn13@gwu.edu], Robel Kahsay [rykahsay@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/pipeline/extractor/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/ln2data/releases/data/v-1.8.24/source/uniprot-proteome-mus-musculus.nt\n Name: mouse_protein_masterlist.csv\n URI: https://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_masterlist.csv\n Output Subdomain\n Name: mouse_protein_submittednames.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_submittednames.csv\n Name: mouse_protein_submittednames.stat.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_submittednames.stat.csv\n\nFields\n Name: \"src_xref_key\"\n Description: \"Source Xref key\"\n Unique Values: \"1\"\n Name: \"xref_id\"\n Description: \"ID of the X-ref\"\n Unique Values: \"119\"\n Name: \"title\"\n Description: \"Title of the published article in the journal\"\n Unique Values: \"119\"\n Name: \"src_xref_id\"\n Description: \"Source X-ref ID\"\n Unique Values: \"120\"\n Name: \"journal_name\"\n Description: \"The name of the journal in which the article is published\"\n Unique Values: \"49\"\n Name: \"authors\"\n Description: \"NA\"\n Unique Values: \"118\"\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: \"117\"\n Name: \"glytoucan_ac\"\n Description: \"Unique accession assigned to the registered glycan structure in GlyTouCan database\"\n Unique Values: \"115\"\n Name: \"xref_key\"\n Description: \"Source key associated to the xref_id\"\n Unique Values: \"1\"", "downloadurl": "/ln2data/releases/data/v-1.8.24/reviewed/mouse_protein_submittednames.csv" }