{ "bcoid": "GLY_000397", "title": "Rat Protein Submitted Names (UniProtKB)", "description": "The Rat Protein Submitted Names (UniProtKB) dataset contains UniProtKB submitted names for rat [taxid:10116] ...", "filename": "rat_protein_submittednames.csv", "filetype": "csv", "categories": { "species": "Rattus norvegicus", "molecule": "Protein", "tags": "Protein Submitted Names", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "D3ZZY8-1...", "Olfactory re..." ], [ "D3ZG10-1...", "GRB2-related..." ] ], "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" } ], [ "D3ZZY8-1", "Olfactory receptor 225", "", "" ], [ "D3ZG10-1", "GRB2-related adaptor protein 2", "", "" ], [ "D3ZB44-1", "Zinc finger protein 324", "", "" ], [ "D3ZB44-1", "RCG27445", "", "" ], [ "D3ZZY2-1", "UTP14A small subunit processome component", "", "" ], [ "Q5RJK5-1", "Chromobox homolog 3 (HP1 gamma homolog, Drosophila), isoform CRA_a", "", "" ], [ "Q5RJK5-1", "Chromobox homolog 3 (HP1 gamma homolog, Drosophila)", "", "" ], [ "Q5RJK5-1", "Chromobox 3", "", "" ], [ "D3ZB48-1", "Hook microtubule-tethering protein 1", "", "" ], [ "D3ZB48-1", "Hook homolog 1 (Drosophila) (Predicted)", "", "" ], [ "Q5RJK6-1", "Inositol polyphosphate-1-phosphatase", "", "" ], [ "Q5RJK6-1", "RCG22563", "", "" ], [ "F1M2W7-1", "Peptidase inhibitor 15", "", "" ], [ "F1M2W7-1", "Protease inhibitor 15 (Predicted)", "", "" ], [ "D4AAV0-1", "Interferon regulatory factor 6 (Predicted)", "", "" ], [ "D4AAV0-1", "Interferon regulatory factor 6", "", "" ], [ "M0RAP1-1", "Undifferentiated embryonic cell transcription factor 1", "", "" ], [ "M0RAP1-1", "RCG47962", "", "" ], [ "M0RAP6-1", "Transcription activation suppressor", "", "" ], [ "A0A0G2JVG5-1", "Trichohyalin-like", "", "" ], [ "A0A0G2JVG6-1", "Uncharacterized protein", "", "" ], [ "M0RAP5-1", "SET-binding factor 1", "", "" ], [ "F1M1W4-1", "Nedd4 family-interacting protein 2", "", "" ], [ "M0R5G7-1", "Keratin-associated protein 22-2", "", "" ], [ "M0R5G5-1", "Polycystin 1-like 2", "", "" ], [ "A0A0G2K1J7-1", "Uncharacterized protein", "", "" ], [ "D3ZZY1-1", "Uncharacterized protein", "", "" ], [ "D3ZLV5-1", "AT-hook transcription factor (Predicted)", "", "" ], [ "D3ZLV5-1", "AT-hook transcription factor", "", "" ], [ "A0A096MJY8-1", "Acetyl-CoA acetyltransferase 2-like 1", "", "" ], [ "D3ZLV9-1", "Kelch-like 23 (Drosophila) (Predicted)", "", "" ], [ "D3ZLV9-1", "Kelch-like family member 23", "", "" ], [ "A0A096MJY1-1", "Similar to Gpc6 protein (Predicted)", "", "" ], [ "A0A096MJY1-1", "Glypican 6", "", "" ], [ "A0A096MJY2-1", "Stomatin-like 3", "", "" ], [ "A0A096MJY2-1", "Stomatin (Epb7.2)-like 3 (Predicted), isoform CRA_a", "", "" ], [ "A0A0G2K5R6-1", "Killer cell lectin-like receptor C2", "", "" ], [ "D4A831-1", "Adhesion G protein-coupled receptor B3", "", "" ], [ "D4A831-1", "Brain-specific angiogenesis inhibitor 3 (Predicted)", "", "" ], [ "D4A832-1", "Solute carrier family 43, member 3", "", "" ], [ "D3ZZY5-1", "P53-induced protein with a death domain-like", "", "" ], [ "D4A834-1", "Proteoglycan 3, pro eosinophil major basic protein 2", "", "" ], [ "D4A834-1", "Proteoglycan 3 (Predicted)", "", "" ], [ "M0R6T1-1", "TatD DNase domain-containing 1", "", "" ], [ "A0A0G2JZ74-1", "Uncharacterized protein", "", "" ], [ "B0JYS4-1", "RCG63627, isoform CRA_a", "", "" ], [ "B0JYS4-1", "Mapk1ip1 protein", "", "" ], [ "B0JYS4-1", "Mitogen-activated protein kinase 1-interacting protein 1", "", "" ], [ "A0A0G2K5R4-1", "Zinc finger, C3H1-type-containing", "", "" ], [ "Q6QI33-1", "RCG65921", "", "" ], [ "Q6QI33-1", "LRRGT00175", "", "" ], [ "M0RCZ7-1", "Antileukoproteinase-like", "", "" ], [ "M0RCZ2-1", "Golgi-associated, olfactory-signaling regulator", "", "" ], [ "M0R7U6-1", "RCG63233", "", "" ], [ "M0R7U6-1", "Sosondowah ankyrin repeat domain family member C", "", "" ], [ "F1M4Y5-1", "Storkhead box 1", "", "" ], [ "F7FL53-1", "Solute carrier family 43 member 1", "", "" ], [ "D3ZNX2-1", "Kinetochore scaffold 1", "", "" ], [ "D4A4W8-1", "T, brachyury homolog (Mouse) (Predicted), isoform CRA_b", "", "" ], [ "D4A4W8-1", "T-box transcription factor T", "", "" ], [ "D4A4W1-1", "Family with sequence similarity 174, member B", "", "" ], [ "A0A0G2K2F3-1", "Keratin-associated protein 17-1", "", "" ], [ "D4A4W4-1", "Abhydrolase domain containing 9 (Predicted)", "", "" ], [ "D4A4W4-1", "Epoxide hydrolase 3", "", "" ], [ "D4A4W6-1", "SRA stem-loop-interacting RNA-binding protein", "", "" ], [ "D4A4W6-1", "RCG20695, isoform CRA_b", "", "" ], [ "D4A4W7-1", "Mitochondria-localized glutamic acid-rich protein", "", "" ], [ "D4A4W7-1", "RCG50000, isoform CRA_c", "", "" ], [ "G3V6I3-1", "Coiled-coil domain containing 49 (Predicted)", "", "" ], [ "G3V6I3-1", "CWC25 spliceosome-associated protein homolog", "", "" ], [ "G3V6I6-1", "Adiponectin receptor 1", "", "" ], [ "G3V6I6-1", "Adiponectin receptor 1, isoform CRA_a", "", "" ], [ "G3V6I4-1", "Mitochondrial amidoxime reducing component 1", "", "" ], [ "G3V6I4-1", "RCG20363, isoform CRA_a", "", "" ], [ "M0R9X3-1", "RNA-binding motif 31, Y-linked", "", "" ], [ "D3ZDS4-1", "EMG1 nucleolar protein homolog (S. cerevisiae) (Predicted), isoform CRA_b", "", "" ], [ "D3ZDS4-1", "EMG1 N1-specific pseudouridine methyltransferase", "", "" ], [ "M0R9X6-1", "Similar to paired immunoglobin-like type 2 receptor beta", "", "" ], [ "M0R9X4-1", "Zinc finger, ZZ-type-containing 3", "", "" ], [ "D3ZVN9-1", "RasGEF domain family, member 1C (Predicted)", "", "" ], [ "D3ZVN9-1", "RasGEF domain family, member 1C", "", "" ], [ "D3ZDS8-1", "Zinc finger RANBP2-type-containing 3", "", "" ], [ "D3ZDS9-1", "Interferon (Alpha and beta) receptor 1 (Predicted), isoform CRA_a", "", "" ], [ "D3ZDS9-1", "Interferon alpha and beta receptor subunit 1", "", "" ] ] }, "readme": "Name\n - Rat Protein Submitted Names (UniProtKB)\n\nVersion\n - 1.8.25\n\nCreated on\n - 2020-03-15T13:43:53.106298\n\nModified on\n - 2020-03-16T11:16:17.615859\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 Rat Protein Submitted Names (UniProtKB) dataset contains UniProtKB submitted names for rat [taxid:10116] 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-rattus-norvegicus.nt\n URI: https://data.glygen.org/ln2data/releases/data/v-1.8.25/source/uniprot-proteome-rattus-norvegicus.nt\n Name: rat_protein_masterlist.csv\n URI: https://data.glygen.org/ln2data/releases/data/v-1.8.25/reviewed/rat_protein_masterlist.csv\n Output Subdomain\n Name: rat_protein_submittednames.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.8.25/reviewed/rat_protein_submittednames.csv\n Name: rat_protein_submittednames.stat.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.8.25/reviewed/rat_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-1.8.25/reviewed/rat_protein_submittednames.csv" }