{ "bcoid": "GLY_000007", "title": "Mouse Proteome Masterlist (UniProtKB)", "description": "The Mouse Proteome Masterlist dataset contains mouse [taxid:10090] UniProtKB canonical accessions mapped ...", "filename": "mouse_protein_masterlist.csv", "filetype": "csv", "categories": { "species": "Mus musculus", "molecule": "Protein", "tags": "Protein Canonical Accessions", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "Q8BH24-1...", "reviewed..." ], [ "Q8BH27-1...", "reviewed..." ] ], "headers": [ "uniprotkb_ca...", "status..." ], "colwidth": [ "40%", "20%" ] }, "sampledata": { "type": "table", "data": [ [ { "type": "string", "label": "uniprotkb_canonical_ac" }, { "type": "string", "label": "status" }, { "type": "string", "label": "gene_name" }, { "type": "string", "label": "reviewed_isoforms" }, { "type": "string", "label": "unreviewed_isoforms" } ], [ "Q8BH24-1", "reviewed", "Tm9sf4", "Q8BH24-1", "" ], [ "Q8BH27-1", "reviewed", "Megf9", "Q8BH27-1", "F8WHT8-1" ], [ "Q8BH26-1", "reviewed", "Tmem251", "Q8BH26-1", "" ], [ "Q8VF69-1", "unreviewed", "Olfr223", "", "Q8VF69-1" ], [ "Q9JHR7-1", "reviewed", "Ide", "Q9JHR7-1", "F6RPJ9-1" ], [ "Q8VF63-1", "unreviewed", "Olfr881", "", "Q7TRE2-1" ], [ "Q8VF63-1", "unreviewed", "Olfr881", "", "Q8VF63-1" ], [ "Q8VF66-1", "reviewed", "Olfr469", "Q8VF66-1", "" ], [ "Q8VF65-1", "reviewed", "Olfr470", "Q8VF65-1", "" ], [ "Q8VF64-1", "unreviewed", "Olfr883", "", "Q8VF64-1" ], [ "B2RXV4-1", "reviewed", "Flvcr1", "B2RXV4-1", "A0A0A6YWJ7-1" ], [ "B2RXV4-1", "reviewed", "Flvcr1", "B2RXV4-2", "A0A0A6YXG8-1" ], [ "E9Q3P4-1", "unreviewed", "Cenpf", "", "E9Q3P4-1" ], [ "E9Q3P4-1", "unreviewed", "Cenpf", "", "E9Q776-1" ], [ "E9Q3P4-1", "unreviewed", "Cenpf", "", "F6V6F0-1" ], [ "E9Q3P4-1", "unreviewed", "Cenpf", "", "E9PY14-1" ], [ "A0A087WRQ3-1", "unreviewed", "Trav1", "", "A0A075B5Z4-1" ], [ "A0A087WRQ3-1", "unreviewed", "Trav1", "", "A0A087WRQ3-1" ], [ "Q9CXG9-1", "reviewed", "Phf19", "Q9CXG9-1", "A0A0J9YU15-1" ], [ "Q07968-1", "reviewed", "F13b", "Q07968-1", "" ], [ "A5H0M4-1", "unreviewed", "Poteg", "", "A5H0M4-1" ], [ "Q9CXG3-1", "reviewed", "Ppil4", "Q9CXG3-1", "D6RJ77-1" ], [ "B1AQ89-1", "unreviewed", "Gm11555", "", "B1AQ89-1" ], [ "B1AQ89-1", "unreviewed", "Gm11555", "", "E9PUW3-1" ], [ "B1AQ87-1", "unreviewed", "Gm14190", "", "B1AQ87-1" ], [ "B1AQ85-1", "unreviewed", "Krtap4-2", "", "B1AQ85-1" ], [ "A0A1B0GST0-1", "unreviewed", "Olfr566", "", "A0A1B0GST0-1" ], [ "A0A1B0GST0-1", "unreviewed", "Olfr566", "", "Q7TRS5-1" ], [ "Q3UU21-1", "unreviewed", "Gm7534", "", "Q3UU21-1" ], [ "P01942-1", "reviewed", "Hba", "P01942-1", "Q91VB8-1" ], [ "P03899-1", "reviewed", "Mtnd3", "P03899-1", "" ], [ "O35153-1", "reviewed", "Bet1l", "O35153-1", "A0A1B0GSF3-1" ], [ "O35153-1", "reviewed", "Bet1l", "", "A0A1B0GR52-1" ], [ "O35153-1", "reviewed", "Bet1l", "", "A0A1B0GQV3-1" ], [ "O35153-1", "reviewed", "Bet1l", "", "A0A1B0GR53-1" ], [ "O35153-1", "reviewed", "Bet1l", "", "A0A1B0GS34-1" ], [ "Q9QYH9-1", "reviewed", "Tnfsf14", "Q9QYH9-1", "" ], [ "P52479-1", "reviewed", "Usp10", "P52479-1", "G3X9U6-1" ], [ "P52479-1", "reviewed", "Usp10", "P52479-2", "G3UYQ8-1" ], [ "B1AXU1-1", "unreviewed", "Gm6812", "", "B1AXU1-1" ], [ "P02762-1", "reviewed", "Mup6", "P02762-1", "A9C496-1" ], [ "P02762-1", "reviewed", "Mup6", "", "A9C497-1" ], [ "Q9QYH6-1", "reviewed", "Maged1", "Q9QYH6-1", "" ], [ "Q3SXJ3-1", "unreviewed", "Olfr1209", "", "Q3SXJ3-1" ], [ "Q08AU9-1", "reviewed", "", "Q08AU9-1", "A0A1B0GR98-1" ], [ "Q08AU9-1", "reviewed", "", "", "A0A1B0GT03-1" ], [ "Q08AU7-1", "reviewed", "Myadml2", "Q08AU7-1", "" ], [ "Q08AU7-1", "reviewed", "Myadml2", "Q08AU7-2", "" ], [ "Q5EBH1-1", "reviewed", "Rassf5", "Q5EBH1-1", "D3Z6G2-1" ], [ "Q5EBH1-1", "reviewed", "Rassf5", "Q5EBH1-2", "A0A1D5RM65-1" ], [ "A0A087WSM5-1", "unreviewed", "Gm29049", "", "A0A087WSM5-1" ], [ "Q61409-1", "reviewed", "Pde3b", "Q61409-1", "E9QLQ3-1" ], [ "A2AVK5-1", "unreviewed", "Olfr1234", "", "A2AVK5-1" ], [ "Q80YU0-1", "reviewed", "Abhd16b", "Q80YU0-1", "" ], [ "Q61400-1", "reviewed", "Ceacam10", "Q61400-1", "A0A0R4J0G1-1" ], [ "Q61400-1", "reviewed", "Ceacam10", "", "A0A0U1RQ24-1" ], [ "Q61400-1", "reviewed", "Ceacam10", "", "A0A0U1RQ82-1" ], [ "Q61406-1", "unreviewed", "Cfhr1", "", "A0A0A6YXP6-1" ], [ "Q61406-1", "unreviewed", "Cfhr1", "", "Q61406-1" ], [ "Q5SRT8-1", "reviewed", "Ccnjl", "Q5SRT8-1", "" ], [ "A0A140LIA7-1", "reviewed", "Gm29797", "A0A140LIA7-1", "" ], [ "P70419-1", "reviewed", "Galnt3", "P70419-1", "" ], [ "Q9EP78-1", "reviewed", "Chst7", "Q9EP78-1", "" ], [ "Q9EP79-1", "reviewed", "Vmn1r52", "Q9EP79-1", "A0A2I3BQJ2-1" ], [ "Q9EP79-1", "reviewed", "Vmn1r52", "", "A0A2I3BQW2-1" ], [ "Q9EP79-1", "reviewed", "Vmn1r52", "", "A0A2I3BRC4-1" ], [ "Q3URK3-1", "reviewed", "Tet1", "Q3URK3-1", "G3UXI7-1" ], [ "Q3URK3-1", "reviewed", "Tet1", "", "G3UZ35-1" ], [ "Q3URK3-1", "reviewed", "Tet1", "", "E9Q9Y4-1" ], [ "Q3URK3-1", "reviewed", "Tet1", "", "G3UZN8-1" ], [ "Q3URK3-1", "reviewed", "Tet1", "", "A0A1W2P7E3-1" ], [ "Q9EP73-1", "reviewed", "Cd274", "Q9EP73-1", "" ], [ "Q9EP71-1", "reviewed", "Rai14", "Q9EP71-1", "A0A2I3BPY7-1" ], [ "Q9EP75-1", "reviewed", "Cyp4f14", "Q9EP75-1", "A0A3Q4EC89-1" ], [ "Q9EP75-1", "reviewed", "Cyp4f14", "", "A0A3Q4EHX3-1" ], [ "Q3URK1-1", "reviewed", "Ccdc190", "Q3URK1-1", "H3BLK0-1" ], [ "Q8BYZ1-1", "reviewed", "Abi3", "Q8BYZ1-1", "F6RXN4-1" ], [ "Q8BYZ1-1", "reviewed", "Abi3", "Q8BYZ1-2", "" ], [ "S4R213-1", "unreviewed", "Lactbl1", "", "S4R213-1" ], [ "P53702-1", "reviewed", "Hccs", "P53702-1", "G3UXK7-1" ], [ "Q8R349-1", "reviewed", "Cdc16", "Q8R349-1", "A0A1B0GR68-1" ], [ "Q8R349-1", "reviewed", "Cdc16", "", "Q3TZ92-1" ], [ "Q8R344-1", "reviewed", "Ccdc12", "Q8R344-1", "A0A0G2JFK6-1" ], [ "Q8R344-1", "reviewed", "Ccdc12", "", "A0A0G2JF16-1" ] ] }, "readme": "Name\n - Mouse Proteome Masterlist (UniProtKB)\n\nVersion\n - 1.5.30\n\nCreated on\n - 2019-01-23T17:21:05.867420\n\nModified on\n - 2020-04-10T12:31:16.220600\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]\n\nAuthors\n - Maria Martin [martin@ebi.ac.uk], Preethi Vasudev [pvasudev@ebi.ac.uk]\n\nChecksum\n - \n\nUsability Domain\n - The Mouse Proteome Masterlist dataset contains mouse [taxid:10090] UniProtKB canonical accessions mapped to the reviewed (SwissProt) and unreviewed (TrEMBL) UniProtKB isoforms via gene grouping. The dataset is derived from 2019-09 UniProtKB release. The dataset is contributed by EMBL-EBI-UniProtKB and processed by GlyGen. If you use this dataset please provide proper attribution to EMBL-EBI-UniProtKB and GlyGen\n\nDescription Domain\n Keywords\n - canonical accessions\n - gene name\n - protein isoforms\n - complete proteome\n - UniProtKB\n\n Pipeline Steps\n - Step-1: The script loads the UniProtKB proteome data from the triplestore containing triples from nt file to map UniProtKB canonical accessions to reviewed and unreviewed isoforms via gene grouping\n\nExecution Domain\n Scripts - https://github.com/GW-HIVE/glygen-backend-integration/blob/master/pipeline/integrator/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: uniprot-proteome-mus-musculus.nt\n URI: https://data.glygen.org/ln2data/releases/data/v-1.5.30/source/uniprot-proteome-mus-musculus.nt\n Output Subdomain\n Name: mouse_protein_masterlist.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.5.30/reviewed/mouse_protein_masterlist.csv\n Name: mouse_protein_masterlist.stat.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.5.30/reviewed/mouse_protein_masterlist.stat.csv\n\nFields\n Name: \"status\"\n Description: \"Status of the protein record in UniProtKB database that can be reviewed for UniProtKB/Swiss-Prot entry or unreviewed for UniProtKB/TrEMBL entry\"\n Unique Values: \"2\"\n Name: \"uniprotkb_canonical_ac\"\n Description: \"Accession assigned to the protein isoform chosen to be the canonical sequence to which all positional annotation refers in UniProtKB database\"\n Unique Values: \"21959\"\n Name: \"unreviewed_isoforms\"\n Description: \"Protein records from UniProtKB/TrEMBL database that have been computationally annotated\"\n Unique Values: \"37599\"\n Name: \"gene_name\"\n Description: \"Official gene name/symbol assigned to the gene. Human gene names/symbols are approved by the HUGO Gene Nomenclature Committee (HGNC)\n Unique Values: \"21605\"\n Name: \"reviewed_isoforms\"\n Description: \"Protein records from UniProtKB/SwissProt database that have been manually curated and annotated based on experimental results\n Unique Values: \"25273\"", "downloadurl": "/ln2data/releases/data/v-1.5.30/reviewed/mouse_protein_masterlist.csv" }