{ "bcoid": "GLY_000022", "title": "Mouse Protein Information (NCBI RefSeq)", "description": "The Mouse Protein Information (NCBI RefSeq) dataset contains mouse [taxid:10090] protein information ...", "filename": "mouse_protein_info_refseq.csv", "filetype": "csv", "categories": { "species": "Mus musculus", "molecule": "Protein", "tags": "Protein Information", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "P03888-1...", "NP_904328..." ], [ "P03893-1...", "NP_904329..." ] ], "headers": [ "uniprotkb_ca...", "p_refseq_ac_..." ], "colwidth": [ "40%", "20%" ] }, "sampledata": { "type": "table", "data": [ [ { "type": "string", "label": "uniprotkb_canonical_ac" }, { "type": "string", "label": "p_refseq_ac_best_match" }, { "type": "string", "label": "refseq_protein_name" }, { "type": "string", "label": "refseq_protein_length" }, { "type": "string", "label": "refseq_protein_summary" } ], [ "P03888-1", "NP_904328", "NADH dehydrogenase subunit 1", "318", "" ], [ "P03893-1", "NP_904329", "NADH dehydrogenase subunit 2", "345", "" ], [ "P00397-1", "NP_904330", "cytochrome c oxidase subunit I", "514", "" ], [ "P00405-1", "NP_904331", "cytochrome c oxidase subunit II", "227", "" ], [ "P03930-1", "NP_904332", "ATP synthase F0 subunit 8", "67", "" ], [ "P00848-1", "NP_904333", "ATP synthase F0 subunit 6", "226", "" ], [ "P00416-1", "NP_904334", "cytochrome c oxidase subunit III", "261", "" ], [ "P03899-1", "NP_904335", "NADH dehydrogenase subunit 3", "115", "" ], [ "P03903-1", "NP_904336", "NADH dehydrogenase subunit 4L", "98", "" ], [ "P03911-1", "NP_904337", "NADH dehydrogenase subunit 4", "459", "" ], [ "P03921-1", "NP_904338", "NADH dehydrogenase subunit 5", "607", "" ], [ "P03925-1", "NP_904339", "NADH dehydrogenase subunit 6", "172", "" ], [ "P00158-1", "NP_904340", "cytochrome b", "381", "" ], [ "Q9CZP3-1", "NP_001077427", "zinc finger protein 655 isoform b", "181", "" ], [ "Q9DCP4-1", "NP_001153110", "uncharacterized protein LOC74088 isoform 2", "199", "" ], [ "A2ANY3-1", "NP_001162062", "uncharacterized protein LOC67715 isoform 1 precursor", "279", "" ], [ "E9Q492-1", "NP_001012311", "piggyBac transposable element-derived protein 1 isoform b", "378", "" ], [ "J3KMM7-1", "NP_001230020", "uncharacterized protein LOC381272 isoform 2", "78", "" ], [ "Q3V0S8-1", "NP_001155900", "uncharacterized protein LOC69324 isoform 1", "276", "" ], [ "Q8BNV8-1", "NP_001107028", "triggering receptor expressed in myeloid cells 4 isoform b precursor", "160", "" ], [ "Q8K193-1", "NP_001297581", "claudin 34C1 isoform 2", "257", "" ], [ "Q80ZX2-1", "NP_001139352", "zinc finger protein 790", "779", "" ], [ "Q8VI97-1", "NP_001153406", "2'-5' oligoadenylate synthetase 1H isoform 2", "239", "" ], [ "Q8BV16-1", "NP_766622", "zinc finger protein 398 isoform 2", "511", "" ], [ "Q8C7I4-1", "NP_001162131", "apolipoprotein L 9b", "310", "" ], [ "E9PYY6-1", "NP_001106859", "sodium-dependent glucose transporter 1 isoform c", "156", "" ], [ "Q9D1U5-1", "NP_001294953", "cutA divalent cation tolerance homolog-like isoform 2", "111", "" ], [ "A0A1B0GT64-1", "NP_001019797", "zinc finger protein 873 isoform 2", "630", "" ], [ "B2RXC8-1", "NP_742156", "serine/threonine-protein phosphatase 2A regulatory subunit B'' subunit alpha isoform 2", "530", "" ], [ "Q24JP3-1", "NP_958753", "protein FAM178B isoform C", "379", "" ], [ "E9Q9Y3-1", "NP_001161841", "uncharacterized protein LOC75453 isoform 1", "381", "" ], [ "E9Q9Y3-1", "NP_083568", "uncharacterized protein LOC75453 isoform 2", "183", "" ], [ "A2AJB9-1", "NP_001263377", "epididymal-specific lipocalin-6 isoform 1 precursor", "245", "" ], [ "Q9CQ47-1", "NP_080009", "uncharacterized protein LOC66732", "274", "" ], [ "A0A571BEV3-1", "NP_083506", "glutamate-rich protein 6B isoform 2", "220", "" ], [ "E9PY36-1", "NP_001074250", "uncharacterized protein C13orf46 homolog isoform 2", "219", "" ], [ "Q24JP3-1", "NP_001119518", "protein FAM178B isoform A", "523", "" ], [ "E9PUA5-1", "NP_444403", "kinesin family member C5B", "672", "" ], [ "J3QNX6-1", "NP_001177947", "keratin associated protein 22-2", "67", "" ], [ "Q9DB24-1", "NP_079631", "transcription elongation factor A (SII)-like 6", "200", "" ], [ "A0A0U1RPC5-1", "NP_064658", "zinc finger protein 109", "644", "" ], [ "Q5GLZ0-1", "NP_001011707", "cytochrome P450, family 2, subfamily c, polypeptide 66 precursor", "490", "" ], [ "Q9CR99-1", "NP_001155245", "ras homolog gene family, member A-like", "193", "" ], [ "G5E8J6-1", "NP_034603", "sarcoplasmic reticulum histidine-rich calcium-binding protein precursor", "738", "" ], [ "E9PVZ2-1", "NP_001157088", "transmembrane protein 273 precursor", "95", "" ], [ "Q9CPT7-1", "NP_080539", "uncharacterized protein LOC67593", "205", "" ], [ "Q0VF76-1", "NP_080627", "tetratricopeptide repeat domain 39D", "604", "" ], [ "Q14BX6-1", "NP_080022", "protein phosphatase 1, regulatory (inhibitor) subunit 14B-like", "153", "" ], [ "D3Z393-1", "NP_001094901", "MKRN2 opposite strand protein", "217", "" ], [ "Q810M4-1", "NP_081582", "zinc finger, DHHC domain containing 25", "279", "" ], [ "Q9D6S9-1", "NP_081431", "uncharacterized protein LOC69661", "174", "" ], [ "Q3V4B7-1", "NP_081110", "keratin associated protein 4-6", "195", "" ], [ "B9EJP6-1", "NP_080670", "late cornified envelope 1F", "147", "" ], [ "Q9CPU1-1", "NP_080768", "synovial sarcoma, X member B1", "170", "" ], [ "K7N6B6-1", "NP_001160314", "vomeronasal 1 receptor 168", "309", "" ], [ "D3YTY2-1", "NP_001160218", "vomeronasal 1 receptor 111", "306", "" ], [ "Q6IMH7-1", "NP_001009545", "a disintegrin and metalloproteinase domain 6-like", "756", "" ], [ "J3KMI7-1", "NP_001186260", "uncharacterized protein LOC100040022", "228", "" ], [ "E9Q459-1", "NP_663458", "zinc finger protein 160", "650", "" ], [ "K7N6B7-1", "NP_001160632", "vomeronasal 1 receptor Vmn1r150", "321", "" ], [ "Q3V490-1", "NP_001032841", "beta-defensin 27 precursor", "66", "" ], [ "Q80YT3-1", "NP_941023", "T-complex protein 1 subunit theta-like 2", "562", "" ], [ "L7N203-1", "NP_001074874", "vomeronasal 2, receptor 28 precursor", "852", "" ], [ "B1B0X6-1", "NP_001076435", "ovary testis transcribed", "433", "" ], [ "E9Q373-1", "NP_001161011", "vomeronasal 1 receptor, 238", "306", "" ], [ "Q8C4L8-1", "NP_001318108", "uncharacterized protein LOC242574", "201", "" ], [ "B2RUQ2-1", "NP_001025060", "basic helix-loop-helix domain-containing protein USF3", "2221", "" ], [ "G3UWI8-1", "NP_001160188", "vomeronasal 1 receptor 113", "307", "" ], [ "D3Z5T8-1", "NP_001159947", "uncharacterized protein C22orf46 homolog", "1512", "" ], [ "E9Q9N3-1", "NP_001160224", "vomeronasal 1 receptor 152", "307", "" ], [ "Q3TUY3-1", "NP_001070889", "spermine binding protein like precursor", "232", "" ], [ "Q9D9M9-1", "NP_083118", "E3 ubiquitin-protein ligase RNF138-like", "248", "" ], [ "A0A0U1RPT6-1", "NP_997131", "uncharacterized protein LOC243944", "632", "" ], [ "L7N226-1", "NP_001160309", "vomeronasal 1 receptor 114", "321", "" ], [ "Q3UX66-1", "NP_001029155", "uncharacterized protein LOC546157", "186", "" ], [ "Q8R285-1", "NP_001160307", "vomeronasal 1 receptor 79", "306", "" ], [ "K7N6J6-1", "NP_001160187", "vomeronasal 1 receptor 120", "305", "" ], [ "J3QP08-1", "NP_001079006", "histone cluster 2 family member", "105", "" ], [ "L7N257-1", "NP_001001332", "stefin A1-like protein", "96", "" ], [ "Q6P8N9-1", "NP_076112", "cytochrome P450, family 2, subfamily d, polypeptide 40", "338", "" ], [ "J3QQ32-1", "NP_001243103", "uncharacterized protein LOC319929", "107", "" ], [ "G3UW31-1", "NP_001160193", "vomeronasal 1 receptor 176", "304", "" ], [ "L7N275-1", "NP_001160315", "vomeronasal 1 receptor 169", "304", "" ], [ "E9PWK2-1", "NP_001160207", "vomeronasal type-1 receptor 2", "298", "" ] ] }, "readme": "Name\n - Mouse Protein Information (NCBI RefSeq)\n\nVersion\n - v-2.0.3\n\nCreated on\n - 2019-01-23T17:21:05.888176\n\nModified on\n - 2022-09-26T08:48:18.844146\n\nCreated by\n - Jeet Vora [jeetvora@gwu.edu], Rahi Navelkar [rsn13@gwu.edu], Robel Kahsay [rykahsay@gwu.edu], Xavier Holmes [xavierh@email.gwu.edu], Karina Martinez [karinamartinez@gwu.edu], Amanda Bell [amandab2140@gwu.edu], Jonathon Keeney [keeneyjg@gwu.edu], Millicent Quartey [mquartey@email.gwu.edu]\n\nAuthors\n - Not Applicable []\n\nChecksum\n - \n\nUsability Domain\n - The Mouse Protein Information (NCBI RefSeq) dataset contains mouse [taxid:10090] protein information such as protein accession, name, length and protein function summary from NCBI RefSeq database. The dataset is derived from NCBI RefSeq Release 96, September 9, 2019. If you use this dataset please provide proper attribution to NCBI RefSeq and GlyGen\n\nDescription Domain\n Keywords\n - protein information\n - protein length\n - protein summary\n - NCBI RefSeq\n\n Pipeline Steps\n - Step-1: The script parses the NCBI RefSeq gpff files to extract the accession, length, protein summary 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: mouse_protein_xref_refseq.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_xref_refseq.csv\n Name: refseq_protein_all_10090.gpff\n URI: /data/shared/glyds/downloads/ncbi/refseq/2022_10/refseq_protein_all_10090.gpff\n Output Subdomain\n Name: mouse_protein_info_refseq.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_info_refseq.csv\n Name: mouse_protein_info_refseq.stat.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.0.3/reviewed/mouse_protein_info_refseq.stat.csv\n\nFields\n Name: \"uniprotkb_canonical_ac\"\n Description: \"\"\n Unique Values: \"20221\"\n Name: \"p_refseq_ac_best_match\"\n Description: \"Best match NCBI RefSeq protein accession to UniProtKB canonical accession\"\n Unique Values: \"23568\"\n Name: \"refseq_protein_name\"\n Description: \"Name of the protein in NCBI RefSeq database\"\n Unique Values: \"23433\"\n Name: \"refseq_protein_length\"\n Description: \"Length of the protein in NCBI RefSeq database\"\n Unique Values: \"2210\"\n Name: \"refseq_protein_summary\"\n Description: \"Function of the protein in NCBI RefSeq database\"\n Unique Values: \"1326\"", "downloadurl": "/ln2data/releases/data/v-2.0.3/reviewed/mouse_protein_info_refseq.csv" }