{ "bcoid": "GLY_000393", "title": "Mouse Protein Names (NCBI RefSeq)", "description": "The Mouse Protein Names (NCBI RefSeq) dataset contains mouse [taxid:10090] protein names (primary and ...", "filename": "mouse_protein_proteinnames_refseq.csv", "filetype": "csv", "categories": { "species": "Mus musculus", "molecule": "Protein", "tags": "Protein Names", "file_type": "csv", "status": "Reviewed", "scope": "internal" }, "minitable": { "content": [ [ "Q8BH24-1...", "NP_598608.2..." ], [ "Q8BH24-1...", "NP_598608.2..." ] ], "headers": [ "uniprotkb_ca...", "refseq_ac..." ], "colwidth": [ "40%", "20%" ] }, "sampledata": { "type": "table", "data": [ [ { "type": "string", "label": "uniprotkb_canonical_ac" }, { "type": "string", "label": "refseq_ac" }, { "type": "string", "label": "refseq_protein_name" } ], [ "Q8BH24-1", "NP_598608.2", "transmembrane 9 superfamily member 4 precursor" ], [ "Q8BH24-1", "NP_598608.2", "transmembrane 9 superfamily member 4" ], [ "Q8BH27-1", "NP_766282.1", "multiple epidermal growth factor-like domains protein 9 precursor" ], [ "Q8BH27-1", "NP_766282.1", "EGF-like-domain, multiple 5; multiple epidermal growth factor-like domains protein 9; EGF-like protein 5; multiple EGF-like domains protein 9; epidermal growth factor-like protein 5" ], [ "Q8BH26-1", "NP_001311444.1", "transmembrane protein 251 isoform 2" ], [ "Q8BH26-1", "NP_001311444.1", "UPF0694 transmembrane protein C14orf109 homolog" ], [ "Q8VF69-1", "NP_666640.1", "olfactory receptor 223" ], [ "Q8VF69-1", "NP_666640.1", "olfactory receptor MOR256-30; novel 7 transmembrane receptor (rhodopsin family) domain-containing protein; GA_x6K02T000SA-78-1037" ], [ "Q8VF63-1", "NP_666530.1", "olfactory receptor 881" ], [ "Q8VF63-1", "NP_666530.1", "olfactory receptor MOR162-7; GA_x6K02T2PVTD-31676771-31677700" ], [ "Q8VF62-1", "NP_666529.1", "olfactory receptor 877" ], [ "Q8VF62-1", "NP_666529.1", "olfactory receptor MOR161-5; GA_x6K02T2PVTD-31540342-31541277" ], [ "Q8VF66-1", "NP_666637.1", "olfactory receptor 469" ], [ "Q8VF66-1", "NP_666637.1", "olfactory receptor MOR204-21; olfactory receptor 204-21; GA_x6K02T2PBJ9-10152980-10152036" ], [ "Q8VF65-1", "NP_666636.1", "olfactory receptor 470" ], [ "Q8VF65-1", "NP_666636.1", "olfactory receptor MOR204-22; olfactory receptor 204-22; GA_x6K02T2PBJ9-10175273-10174329" ], [ "Q8VF64-1", "NP_666531.1", "olfactory receptor 883" ], [ "Q8VF64-1", "NP_666531.1", "olfactory receptor MOR162-6; GA_x6K02T2PVTD-31705144-31706073" ], [ "B2RXV4-1", "NP_001074728.1", "feline leukemia virus subgroup C receptor-related protein 1 isoform 1" ], [ "B2RXV4-1", "NP_001074728.1", "feline leukemia virus subgroup C receptor-related protein 1; major facilitator superfamily domain containing 7B" ], [ "E9Q3P4-1", "NP_001074832.2", "centromere protein F" ], [ "E9Q3P4-1", "NP_001074832.2", "leucine, glutamic acid, lysine family 1 protein; centromere autoantigen F" ], [ "Q9CXG9-1", "NP_082992.1", "PHD finger protein 19" ], [ "Q9CXG9-1", "NP_082992.1", "polycomb-like protein 3" ], [ "Q07968-1", "NP_112441.2", "coagulation factor XIII B chain precursor" ], [ "Q07968-1", "NP_112441.2", "coagulation factor XIII B chain; transglutaminase B chain; protein-glutamine gamma-glutamyltransferase B chain" ], [ "A5H0M4-1", "NP_081894.1", "POTE ankyrin domain family member A isoform 2" ], [ "A5H0M4-1", "NP_081894.1", "gonad specific ankyrin repeat protein" ], [ "Q9CXG3-1", "NP_080417.2", "peptidyl-prolyl cis-trans isomerase-like 4" ], [ "Q9CXG3-1", "NP_080417.2", "cyclophilin-like protein PPIL4; rotamase PPIL4" ], [ "B1AQ85-1", "NP_081083.2", "keratin associated protein 4-2" ], [ "P03899-1", "NP_904335.1", "NADH dehydrogenase subunit 3" ], [ "O35153-1", "NP_061212.3", "BET1-like protein" ], [ "O35153-1", "NP_061212.3", "soluble N-ethylmaleimide-sensitive factor attachment protein (SNARE) 15; GOS-15; vesicle transport protein GOS15; golgi SNARE with a size of 15 kDa; golgi SNARE, 15 kD; bet1l; blocked early in transport 1 homolog-like" ], [ "P03893-1", "NP_904329.1", "NADH dehydrogenase subunit 2" ], [ "Q9QYH9-1", "NP_062291.1", "tumor necrosis factor ligand superfamily member 14" ], [ "Q9QYH9-1", "NP_062291.1", "tumor necrosis factor ligand 1d" ], [ "P52479-1", "NP_001297559.1", "ubiquitin carboxyl-terminal hydrolase 10 isoform 2" ], [ "P52479-1", "NP_001297559.1", "ubiquintin c-terminal hydrolase related polypeptide; ubiquitin specific protease 10; ubiquitin carboxyl-terminal hydrolase 10; ubiquitin thioesterase 10; deubiquitinating enzyme 10; ubiquitin-specific-processing protease 10; ubiquitin thiolesterase 10" ], [ "P02762-1", "NP_001128599.1", "major urinary protein 11 precursor" ], [ "P02762-1", "NP_001128599.1", "major urinary protein 17; major urinary protein 14; Major urinary protein 6; MUP 6; Alpha-2U-globulin; Group 1, BS6; MUP11 and MUP8; Major urinary proteins 11 and 8; major urinary protein 11" ], [ "Q9QYH6-1", "NP_062765.1", "melanoma-associated antigen D1" ], [ "Q9QYH6-1", "NP_062765.1", "MAGE-D1 antigen; neurotrophin receptor-interacting MAGE homolog" ], [ "Q3SXJ3-1", "NP_666672.2", "olfactory receptor 1209" ], [ "Q3SXJ3-1", "NP_666672.2", "olfactory receptor MOR230-7; GA_x6K02T2Q125-50386882-50385950" ], [ "O08537-1", "NP_034287.3", "estrogen receptor beta isoform 2" ], [ "O08537-1", "NP_034287.3", "ER-beta; nuclear receptor subfamily 3 group A member 2; estrogen receptor beta; oestrogen receptor beta" ], [ "Q08AU9-1", "NP_001156945.1", "UPF0450 protein C17orf58 homolog isoform 1 precursor" ], [ "Q08AU9-1", "NP_001156945.1", "UPF0450 protein C17orf58 homolog" ], [ "Q08AU7-1", "NP_001191749.1", "myeloid-associated differentiation marker-like protein 2 isoform 2" ], [ "Q08AU7-1", "NP_001191749.1", "myeloid-associated differentiation marker-like protein 2" ], [ "Q5EBH1-1", "NP_001298023.1", "ras association domain-containing protein 5 isoform 2" ], [ "Q5EBH1-1", "NP_001298023.1", "ras association domain-containing protein 5; Rap1-binding protein; new ras effector 1; regulator for cell adhesion and polarization enriched in lymphoid tissues; novel ras effector 1" ], [ "Q61409-1", "NP_035185.2", "cGMP-inhibited 3',5'-cyclic phosphodiesterase B" ], [ "Q61409-1", "NP_035185.2", "CGIPDE1; CGI-PDE B; cyclic GMP-inhibited phosphodiesterase B" ], [ "A2AVK5-1", "NP_667184.2", "olfactory receptor 1234" ], [ "A2AVK5-1", "NP_667184.2", "olfactory receptor MOR231-2; GA_x6K02T2Q125-50805620-50804676" ], [ "Q80YU0-1", "NP_899004.1", "protein ABHD16B" ], [ "Q80YU0-1", "NP_899004.1", "abhydrolase domain-containing protein 16B; alpha/beta hydrolase domain-containing protein 16B" ], [ "Q61400-1", "NP_031701.3", "carcinoembryonic antigen-related cell adhesion molecule 10 isoform 1 precursor" ], [ "Q61400-1", "NP_031701.3", "CEA-related cell adhesion molecule 10; carcinoembryonic antigen 10" ], [ "Q61406-1", "NP_056595.1", "complement factor H-related 1 precursor" ], [ "Q61406-1", "NP_056595.1", "complement component factor h-like 1; complement factor H-related protein" ], [ "Q5SRT8-1", "NP_001038995.1", "cyclin-J-like protein" ], [ "P70414-1", "NP_035536.2", "sodium/calcium exchanger 1 isoform A precursor" ], [ "P70414-1", "NP_035536.2", "sodium/calcium exchanger 1; Na/Ca exchanger; sodium/calcium exchanger isoform NaCa7; Na(+)/Ca(2+)-exchange protein 1; solute carrier family 8 member 1" ], [ "P70419-1", "NP_056551.2", "polypeptide N-acetylgalactosaminyltransferase 3" ], [ "P70419-1", "NP_056551.2", "galNAc-T3; pp-GaNTase 3; polypeptide GalNAc transferase 3; protein-UDP acetylgalactosaminyltransferase 3; UDP-GalNAc:polypeptide N-acetylgalactosaminyltransferase 3; ppGaNTase-T3; UDP-N-acetyl-alpha-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase 3" ], [ "Q9EP78-1", "NP_068361.1", "carbohydrate sulfotransferase 7 precursor" ], [ "Q9EP78-1", "NP_068361.1", "carbohydrate sulfotransferase 7; chondroitin 6-sulfotransferase 2; N-acetylglucosamine 6-O-sulfotransferase 1; galactose/N-acetylglucosamine/N-acetylglucosamine 6-O-sulfotransferase 5; N-acetylglucosamine 6-O-sulfotransferase 4" ], [ "Q9EP79-1", "NP_444452.1", "vomeronasal type-1 receptor 52" ], [ "Q9EP79-1", "NP_444452.1", "vomeronasal type-1 receptor A7; pheromone receptor VN3; vomeronasal receptor 3; vomeronasal 1 receptor, A7; vomernasal 1 receptor Vmn1r52" ], [ "Q9EP72-1", "NP_598510.1", "ER membrane protein complex subunit 7 precursor" ], [ "Q9EP72-1", "NP_598510.1", "UPF0480 protein C15orf24 homolog" ], [ "Q9EP73-1", "NP_068693.1", "programmed cell death 1 ligand 1 precursor" ], [ "Q9EP73-1", "NP_068693.1", "programmed cell death 1 ligand 1; B7 homolog 1; PDCD1 ligand 1" ], [ "Q9EP71-1", "NP_001159880.1", "ankycorbin isoform 1" ], [ "Q9EP71-1", "NP_001159880.1", "NORPEG-like protein; p125; retinoic acid-induced protein 14; novel retinal pigment epithelial cell protein; ankyrin repeat and coiled-coil structure-containing protein" ], [ "Q9EP75-1", "NP_001191262.1", "leukotriene-B4 omega-hydroxylase 3" ], [ "Q9EP75-1", "NP_001191262.1", "cytochrome P450 4F14; cytochrome P450-LTB-omega; leukotriene-B4 20-monooxygenase 3; cytochrome P450, subfamily IVF, polypeptide 14 (leukotriene B4 omega hydroxylase)" ], [ "Q3URK1-1", "NP_001028357.1", "coiled-coil domain-containing protein 190 isoform 1" ], [ "Q3URK1-1", "NP_001028357.1", "coiled-coil domain-containing protein 190" ], [ "S4R213-1", "NP_001230191.1", "putative beta-lactamase-like 1" ], [ "P53702-1", "NP_001317978.1", "cytochrome c-type heme lyase" ] ] }, "readme": "Name\n - Mouse Protein Names (NCBI RefSeq)\n\nVersion\n - 1.5.18\n\nCreated on\n - 2020-03-15T13:29:07.339002\n\nModified on\n - 2020-03-16T11:18:46.916940\n\nCreated by\n - Jeet Vora [jeetvora@gwu.edu], Rahi Navelkar [rsn13@gwu.edu], Robel Kahsay [rykahsay@gwu.edu]\n\nAuthors\n - \n\nChecksum\n - \n\nUsability Domain\n - The Mouse Protein Names (NCBI RefSeq) dataset contains mouse [taxid:10090] protein names (primary and other alternative/synonym protein names) from NCBI RefSeq database. The dataset is derived from NCBI RefSeq Release 96, September 9, 2019. 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 names\n - NCBI RefSeq\n\n Pipeline Steps\n - Step-1: The script parses the NCBI RefSeq gpff files to extract the protein names and then maps the NCBI RefSeq accesion to the UniProtKB Canonical accessions\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: refseq_protein_all_10090.gpff\n URI: https://data.glygen.org/ln2data/releases/data/v-1.5.18/source/refseq_protein_all_10090.gpff\n Name: mouse_protein_xref_refseq.csv\n URI: https://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/mouse_protein_xref_refseq.csv\n Output Subdomain\n Name: mouse_protein_proteinnames_refseq.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/mouse_protein_proteinnames_refseq.csv\n Name: mouse_protein_proteinnames_refseq.stat.csv\n URI: http://data.glygen.org/ln2data/releases/data/v-1.5.18/reviewed/mouse_protein_proteinnames_refseq.stat.csv\n\nFields\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: \"19422\"\n Name: \"refseq_ac\"\n Description: \"Accession in the NCBI RefSeq database\"\n Unique Values: \"19395\"\n Name: \"refseq_protein_name\"\n Description: \"Name of the protein in NCBI RefSeq database\"\n Unique Values: \"34974\"", "downloadurl": "/ln2data/releases/data/v-1.5.18/reviewed/mouse_protein_proteinnames_refseq.csv" }