{ "bcoid": "GLY_000950", "title": "Pig Proteome Masterlist (UniProtKB)", "description": "The Pig Proteome Materlist dataset contains pig [taxid:9823] UniProtKB canonical accessions mapped to ...", "filename": "pig_protein_masterlist.csv", "filetype": "csv", "categories": { "species": [ "Sus scrofa" ], "molecule": [ "Protein" ], "tag": [ "Protein Canonical Accessions" ], "file_type": [ "csv" ], "status": [ "Reviewed" ], "scope": [ "internal" ] }, "minitable": { "content": [ [ "A0A8W4F999-1...", "unreviewed..." ], [ "A0A287BN22-1...", "unreviewed..." ] ], "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" } ], [ "A0A8W4F999-1", "unreviewed", "", "", "A0A8W4F999-1" ], [ "A0A287BN22-1", "unreviewed", "BTBD10", "", "A0A287A7P5-1" ], [ "A0A287BN22-1", "unreviewed", "BTBD10", "", "A0A287BN22-1" ], [ "A0A287A8B3-1", "unreviewed", "CALCB", "", "A0A287A8B3-1" ], [ "A0A287BN27-1", "unreviewed", "PABIR2", "", "A0A287BN27-1" ], [ "A0A287BN27-1", "unreviewed", "PABIR2", "", "A0A5G2QTE9-1" ], [ "A0A287BN27-1", "unreviewed", "PABIR2", "", "F1RTD5-1" ], [ "A0A287BN27-1", "unreviewed", "PABIR2", "", "I3LPJ1-1" ], [ "A0A5G2QDB0-1", "unreviewed", "", "", "A0A5G2QDB0-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "A0A287A5F7-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "A0A287APB4-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "A0A5G2Q9H1-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "A0A5G2QH01-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "A0A8W4FI60-1" ], [ "A0A287A5F7-1", "unreviewed", "PPP2R1B", "", "I3L7S2-1" ], [ "D7PF45-1", "reviewed", "INPPL1", "D7PF45-1", "" ], [ "A0A4X1U326-1", "unreviewed", "RAB38", "", "A0A4X1U326-1" ], [ "A0A8W4FJE4-1", "unreviewed", "", "", "A0A8W4FJE4-1" ], [ "A0A8W4FJE7-1", "unreviewed", "", "", "A0A8W4FJE7-1" ], [ "A0A8W4FDQ3-1", "unreviewed", "", "", "A0A8W4FDQ3-1" ], [ "A0A287ANF4-1", "unreviewed", "TOGARAM1", "", "A0A287ANF4-1" ], [ "A0A287ANF4-1", "unreviewed", "TOGARAM1", "", "A0A287B4D3-1" ], [ "A0A287ANF4-1", "unreviewed", "TOGARAM1", "", "A0A287BHG3-1" ], [ "A0A287ANF5-1", "unreviewed", "AVEN", "", "A0A287ANF5-1" ], [ "I3LFU4-1", "unreviewed", "CCDC14", "", "I3LFU4-1" ], [ "I3LFU4-1", "unreviewed", "CCDC14", "", "A0A287B8B9-1" ], [ "A0A287ANF8-1", "unreviewed", "ATPAF1", "", "A0A287ANF8-1" ], [ "A0A287ANF8-1", "unreviewed", "ATPAF1", "", "A0A5G2QLQ2-1" ], [ "A0A287ANF8-1", "unreviewed", "ATPAF1", "", "A0A5G2QUN6-1" ], [ "A0A287ANF8-1", "unreviewed", "ATPAF1", "", "A0A8W4FPU7-1" ], [ "A0A287ANF9-1", "unreviewed", "PTGS1", "", "A0A287ANF9-1" ], [ "I3LQS2-1", "unreviewed", "NT5C", "", "I3LQS2-1" ], [ "I3LHI0-1", "unreviewed", "KRT23", "", "I3LHI0-1" ], [ "Q6RVA9-1", "reviewed", "CAV1", "Q6RVA9-1", "" ], [ "P05383-1", "reviewed", "PRKACB", "P05383-1", "" ], [ "F1RTM0-1", "unreviewed", "LOC100522569", "", "A0A8D0ZQE4-1" ], [ "F1RTM0-1", "unreviewed", "LOC100522569", "", "F1RTM0-1" ], [ "F1RTM6-1", "unreviewed", "AR", "", "A0A287BJP3-1" ], [ "F1RTM6-1", "unreviewed", "AR", "", "A0A5G2QRH9-1" ], [ "F1RTM6-1", "unreviewed", "AR", "", "F1RTM6-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "A0A286ZM59-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "A0A287AWG4-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "A0A5G2QM40-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "A0A5G2QSG4-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "A0A5G2RAC2-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "F1RKF7-1" ], [ "A0A5G2QM40-1", "unreviewed", "CUX1", "", "F1RKF9-1" ], [ "F1S7I3-1", "unreviewed", "DEFB115", "", "F1S7I3-1" ], [ "A0A287A8B9-1", "unreviewed", "PATE2", "", "A0A287A8B9-1" ], [ "A0A287A8B9-1", "unreviewed", "PATE2", "", "A0A4X1VG44-1" ], [ "F1SJB9-1", "unreviewed", "CEMIP2", "", "A0A287ACI4-1" ], [ "F1SJB9-1", "unreviewed", "CEMIP2", "", "A0A8D0SB13-1" ], [ "F1SJB9-1", "unreviewed", "CEMIP2", "", "F1SJB9-1" ], [ "A0A8W4FCQ5-1", "unreviewed", "H1-2", "", "A0A8W4FCQ5-1" ], [ "A0A8W4FCQ0-1", "unreviewed", "", "", "A0A8W4FCQ0-1" ], [ "A0A8W4FCQ3-1", "unreviewed", "ANKRD39", "", "A0A8W4FCQ3-1" ], [ "A0A8W4FCQ2-1", "unreviewed", "", "", "A0A8W4FCQ2-1" ], [ "A0A8W4FCQ2-1", "unreviewed", "", "", "A0A8X9AEV2-1" ], [ "A0A287ATT8-1", "unreviewed", "LOC100514973", "", "A0A287ATT8-1" ], [ "A0A287ATT9-1", "unreviewed", "PTTG1IP", "", "A0A287ATT9-1" ], [ "F1SJI8-1", "unreviewed", "C1H15orf61", "", "F1SJI8-1" ], [ "C0JPM4-1", "unreviewed", "TIMP2", "", "A0A5G2QR70-1" ], [ "C0JPM4-1", "unreviewed", "TIMP2", "", "A0A5G2QTA4-1" ], [ "C0JPM4-1", "unreviewed", "TIMP2", "", "A0A5G2R175-1" ], [ "C0JPM4-1", "unreviewed", "TIMP2", "", "C0JPM4-1" ], [ "A0A287B1P5-1", "unreviewed", "GUCD1", "", "A0A287A1F7-1" ], [ "A0A287B1P5-1", "unreviewed", "GUCD1", "", "A0A287B1P5-1" ], [ "A0A287B1P5-1", "unreviewed", "GUCD1", "", "A0A5G2QNG6-1" ], [ "A0A287B1P5-1", "unreviewed", "GUCD1", "", "F1RL42-1" ], [ "F1SJI3-1", "unreviewed", "RBM44", "", "F1SJI3-1" ], [ "F1RMF0-1", "unreviewed", "OR5B21", "", "F1RMF0-1" ], [ "A0A4X1V1L8-1", "unreviewed", "GSTT4", "", "A0A4X1V1L8-1" ], [ "A0A4X1UQU7-1", "unreviewed", "", "", "A0A4X1UQU7-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A5G2QC32-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A5G2QLZ7-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A8W4FJT6-1" ], [ "F1RMF6-1", "unreviewed", "", "", "F1RMF6-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A287A7E5-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A287AJ40-1" ], [ "F1RMF6-1", "unreviewed", "", "", "A0A287AVQ4-1" ], [ "F1RMF7-1", "unreviewed", "GLYAT", "", "F1RMF7-1" ], [ "F1RMF7-1", "unreviewed", "GLYAT", "", "I3LB63-1" ], [ "F1RMF9-1", "unreviewed", "FAM111B", "", "F1RMF9-1" ], [ "Q8MIS1-1", "unreviewed", "KCNE3", "", "Q8MIS1-1" ] ] }, "readme": "Name\n - Pig Proteome Masterlist (UniProtKB)\n\nVersion\n - v-2.4.1\n\nCreated on\n - \n\nModified on\n - 2024-02-28T18:54:42\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], Luke Johnson [luke.johnson@email.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 Pig Proteome Materlist dataset contains pig [taxid:9823] 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. 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/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: uniprot-proteome-sus_scrofa.nt\n URI: https:/data.glygen.org/ln2downloads/ebi/2024_02_13/uniprot-proteome-sus_scrofa.nt\n Output Subdomain\n Name: pig_protein_masterlist.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/pig_protein_masterlist.csv\n Name: pig_protein_masterlist.stat.csv\n URI: https:/data.glygen.org/ln2releases/data/v-2.4.1/reviewed/pig_protein_masterlist.stat.csv\n\nFields\n Name: \"uniprotkb_canonical_ac\"\n Description: \"\"\n Unique Values: \"22786\"\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: \"gene_name\"\n Description: \"\"Official gene name/symbol assigned to the gene. Human gene name/symbol are approved by the HUGO Gene Nomenclature Committee (HGNC) and mouse gene name are approved by the Mouse Genome Informatics (MGI)\"\n Unique Values: \"17959\"\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: \"1534\"\n Name: \"unreviewed_isoforms\"\n Description: \"Protein records from UniProtKB/TrEMBL database that have been computationally annotated\"\n Unique Values: \"44720\"", "downloadurl": "/ln2data/releases/data/v-2.4.1/reviewed/pig_protein_masterlist.csv" }