{
    "bcoid": "GLY_000399",
    "title": "Mouse Protein Sequence Info (UniProtKB)",
    "description": "The Mouse Protein Sequence Info (UniProtKB) dataset contains mouse [taxid:10090] UniProtKB protein fasta ...",
    "filename": "mouse_protein_sequenceinfo.csv",
    "filetype": "csv",
    "categories": {
        "species": "Mus musculus",
        "molecule": "Protein",
        "tags": "All Protein Sequence Info",
        "file_type": "csv",
        "status": "Reviewed",
        "scope": "internal"
    },
    "minitable": {
        "content": [
            [
                "Q8BH24-1...",
                "Q8BH24-1..."
            ],
            [
                "Q8BH27-1...",
                "F8WHT8-1..."
            ]
        ],
        "headers": [
            "uniprotkb_ca...",
            "uniprotkb_is..."
        ],
        "colwidth": [
            "40%",
            "20%"
        ]
    },
    "sampledata": {
        "type": "table",
        "data": [
            [
                {
                    "type": "string",
                    "label": "uniprotkb_canonical_ac"
                },
                {
                    "type": "string",
                    "label": "uniprotkb_isoform_ac"
                },
                {
                    "type": "string",
                    "label": "protein_existence"
                },
                {
                    "type": "string",
                    "label": "sequence_version"
                },
                {
                    "type": "string",
                    "label": "sequence_header"
                }
            ],
            [
                "Q8BH24-1",
                "Q8BH24-1",
                "1",
                "1",
                "sp|Q8BH24-1|TM9S4_MOUSE Isoform 1 of Transmembrane 9 superfamily member 4  OS=Mus musculus OX=10090 GN=Tm9sf4 PE=1 SV=1 CANONICAL=Q8BH24-1"
            ],
            [
                "Q8BH27-1",
                "F8WHT8-1",
                "2",
                "1",
                "tr|F8WHT8-1|MEGF9_MOUSE Computationally mapped isoform of Multiple epidermal growth factor-like domains protein 9  OS=Mus musculus OX=10090 GN=Megf9 PE=2 SV=1 CANONICAL=Q8BH27-1"
            ],
            [
                "Q8BH27-1",
                "Q8BH27-1",
                "2",
                "1",
                "sp|Q8BH27-1|MEGF9_MOUSE Isoform 1 of Multiple epidermal growth factor-like domains protein 9  OS=Mus musculus OX=10090 GN=Megf9 PE=2 SV=1 CANONICAL=Q8BH27-1"
            ],
            [
                "Q8BH26-1",
                "Q8BH26-1",
                "2",
                "1",
                "sp|Q8BH26-1|TM251_MOUSE Isoform 1 of Transmembrane protein 251  OS=Mus musculus OX=10090 GN=Tmem251 PE=2 SV=1 CANONICAL=Q8BH26-1"
            ],
            [
                "Q922G0-1",
                "F6R114-1",
                "1",
                "1",
                "tr|F6R114-1|F6R114_MOUSE Computationally mapped isoform of Solute carrier family 25 member 36  OS=Mus musculus OX=10090 GN=Slc25a36 PE=1 SV=1 CANONICAL=Q922G0-1"
            ],
            [
                "Q9DAV8-1",
                "J3JS25-1",
                "3",
                "1",
                "tr|J3JS25-1|J3JS25_MOUSE Computationally mapped isoform of Prolactin family 8, subfamily a, member 1  OS=Mus musculus OX=10090 GN=Prl8a1 PE=3 SV=1 CANONICAL=Q9DAV8-1"
            ],
            [
                "O35453-1",
                "F7C9T6-1",
                "4",
                "1",
                "tr|F7C9T6-1|F7C9T6_MOUSE Computationally mapped isoform of Serine protease hepsin  OS=Mus musculus OX=10090 GN=Hpn PE=4 SV=1 CANONICAL=O35453-1"
            ],
            [
                "Q8VF69-1",
                "Q8VF69-1",
                "2",
                "1",
                "tr|Q8VF69-1|Q8VF69_MOUSE Isoform 1 of Olfactory receptor  OS=Mus musculus OX=10090 GN=Olfr223 PE=2 SV=1 CANONICAL=Q8VF69-1"
            ],
            [
                "Q9JHR7-1",
                "Q9JHR7-1",
                "1",
                "1",
                "sp|Q9JHR7-1|IDE_MOUSE Isoform 1 of Insulin-degrading enzyme  OS=Mus musculus OX=10090 GN=Ide PE=1 SV=1 CANONICAL=Q9JHR7-1"
            ],
            [
                "Q9JHR7-1",
                "Q8CGB9-1",
                "1",
                "1",
                "tr|Q8CGB9-1|IDE_MOUSE Computationally mapped isoform of Insulin-degrading enzyme  OS=Mus musculus OX=10090 GN=Ide PE=1 SV=1 CANONICAL=Q9JHR7-1"
            ],
            [
                "Q8VF63-1",
                "Q7TRE2-1",
                "4",
                "1",
                "tr|Q7TRE2-1|Q8VF63_MOUSE Computationally mapped isoform of Olfactory receptor 881  OS=Mus musculus OX=10090 GN=Olfr881 PE=4 SV=1 CANONICAL=Q8VF63-1"
            ],
            [
                "Q8VF63-1",
                "Q8VF63-1",
                "4",
                "1",
                "tr|Q8VF63-1|Q8VF63_MOUSE Isoform 1 of Olfactory receptor 881  OS=Mus musculus OX=10090 GN=Olfr881 PE=4 SV=1 CANONICAL=Q8VF63-1"
            ],
            [
                "Q61476-1",
                "E9Q731-1",
                "4",
                "1",
                "tr|E9Q731-1|E9Q731_MOUSE Computationally mapped isoform of Complement decay-accelerating factor transmembrane isoform  OS=Mus musculus OX=10090 GN=Cd55b PE=4 SV=1 CANONICAL=Q61476-1"
            ],
            [
                "Q8VF60-1",
                "Q7TS12-1",
                "3",
                "1",
                "tr|Q7TS12-1|Q8VF60_MOUSE Computationally mapped isoform of Olfactory receptor  OS=Mus musculus OX=10090 GN=Olfr291 PE=3 SV=1 CANONICAL=Q8VF60-1"
            ],
            [
                "Q8VF60-1",
                "Q8VF60-1",
                "3",
                "1",
                "tr|Q8VF60-1|Q8VF60_MOUSE Isoform 1 of Olfactory receptor  OS=Mus musculus OX=10090 GN=Olfr291 PE=3 SV=1 CANONICAL=Q8VF60-1"
            ],
            [
                "Q9Z2D8-1",
                "F6WZH1-1",
                "1",
                "1",
                "tr|F6WZH1-1|F6WZH1_MOUSE Computationally mapped isoform of Methyl-CpG-binding domain protein 3  OS=Mus musculus OX=10090 GN=Mbd3 PE=1 SV=1 CANONICAL=Q9Z2D8-1"
            ],
            [
                "Q8VF66-1",
                "Q8VF66-1",
                "3",
                "1",
                "sp|Q8VF66-1|OL469_MOUSE Isoform 1 of Olfactory receptor 469  OS=Mus musculus OX=10090 GN=Olfr469 PE=3 SV=1 CANONICAL=Q8VF66-1"
            ],
            [
                "Q8VF65-1",
                "Q8VF65-1",
                "3",
                "1",
                "sp|Q8VF65-1|OL470_MOUSE Isoform 1 of Olfactory receptor 470  OS=Mus musculus OX=10090 GN=Olfr470 PE=3 SV=1 CANONICAL=Q8VF65-1"
            ],
            [
                "Q8VF64-1",
                "Q8VF64-1",
                "4",
                "1",
                "tr|Q8VF64-1|Q8VF64_MOUSE Isoform 1 of Olfactory receptor 883  OS=Mus musculus OX=10090 GN=Olfr883 PE=4 SV=1 CANONICAL=Q8VF64-1"
            ],
            [
                "Q80T32-1",
                "B2RXV6-1",
                "2",
                "1",
                "tr|B2RXV6-1|B2RXV6_MOUSE Computationally mapped isoform of Gpr133 protein  OS=Mus musculus OX=10090 GN=Adgrd1 PE=2 SV=1 CANONICAL=Q80T32-1"
            ],
            [
                "B2RXV4-1",
                "B2RXV4-1",
                "2",
                "1",
                "sp|B2RXV4-1|FLVC1_MOUSE Isoform 1 of Feline leukemia virus subgroup C receptor-related protein 1  OS=Mus musculus OX=10090 GN=Flvcr1 PE=2 SV=1 CANONICAL=B2RXV4-1"
            ],
            [
                "B2RXV4-1",
                "B2RXV4-2",
                "2",
                "1",
                "sp|B2RXV4-2|FLVC1_MOUSE Isoform 2 of Feline leukemia virus subgroup C receptor-related protein 1  OS=Mus musculus OX=10090 GN=Flvcr1 PE=2 SV=1 CANONICAL=B2RXV4-1"
            ],
            [
                "B2RXV4-1",
                "A0A0A6YWJ7-1",
                "2",
                "1",
                "tr|A0A0A6YWJ7-1|FLVC1_MOUSE Computationally mapped isoform of Feline leukemia virus subgroup C receptor-related protein 1  OS=Mus musculus OX=10090 GN=Flvcr1 PE=2 SV=1 CANONICAL=B2RXV4-1"
            ],
            [
                "B2RXV4-1",
                "A0A0A6YXG8-1",
                "2",
                "1",
                "tr|A0A0A6YXG8-1|FLVC1_MOUSE Computationally mapped isoform of Feline leukemia virus subgroup C receptor-related protein 1  OS=Mus musculus OX=10090 GN=Flvcr1 PE=2 SV=1 CANONICAL=B2RXV4-1"
            ],
            [
                "P62492-1",
                "E9Q3P9-1",
                "4",
                "1",
                "tr|E9Q3P9-1|E9Q3P9_MOUSE Computationally mapped isoform of Ras-related protein Rab-11A  OS=Mus musculus OX=10090 GN=Rab11a PE=4 SV=1 CANONICAL=P62492-1"
            ],
            [
                "E9Q3P4-1",
                "E9Q3P4-1",
                "1",
                "1",
                "tr|E9Q3P4-1|E9Q3P4_MOUSE Isoform 1 of Centromere protein F  OS=Mus musculus OX=10090 GN=Cenpf PE=1 SV=1 CANONICAL=E9Q3P4-1"
            ],
            [
                "E9Q3P4-1",
                "E9Q776-1",
                "1",
                "1",
                "tr|E9Q776-1|E9Q3P4_MOUSE Computationally mapped isoform of Centromere protein F  OS=Mus musculus OX=10090 GN=Cenpf PE=1 SV=1 CANONICAL=E9Q3P4-1"
            ],
            [
                "E9Q3P4-1",
                "F6V6F0-1",
                "1",
                "1",
                "tr|F6V6F0-1|E9Q3P4_MOUSE Computationally mapped isoform of Centromere protein F  OS=Mus musculus OX=10090 GN=Cenpf PE=1 SV=1 CANONICAL=E9Q3P4-1"
            ],
            [
                "E9Q3P4-1",
                "E9PY14-1",
                "1",
                "1",
                "tr|E9PY14-1|E9Q3P4_MOUSE Computationally mapped isoform of Centromere protein F  OS=Mus musculus OX=10090 GN=Cenpf PE=1 SV=1 CANONICAL=E9Q3P4-1"
            ],
            [
                "P63166-1",
                "A0A087WRQ4-1",
                "1",
                "1",
                "tr|A0A087WRQ4-1|A0A087WRQ4_MOUSE Computationally mapped isoform of Small ubiquitin-related modifier 1  OS=Mus musculus OX=10090 GN=Sumo1 PE=1 SV=1 CANONICAL=P63166-1"
            ],
            [
                "D3YX29-1",
                "A0A087WRQ7-1",
                "4",
                "1",
                "tr|A0A087WRQ7-1|A0A087WRQ7_MOUSE Computationally mapped isoform of Oocyte maturation, beta  OS=Mus musculus OX=10090 GN=Omt2b PE=4 SV=1 CANONICAL=D3YX29-1"
            ],
            [
                "A0A087WPJ1-1",
                "A0A087WRQ1-1",
                "4",
                "1",
                "tr|A0A087WRQ1-1|A0A087WRQ1_MOUSE Computationally mapped isoform of RIKEN cDNA 1700017N19 gene  OS=Mus musculus OX=10090 GN=1700017N19Rik PE=4 SV=1 CANONICAL=A0A087WPJ1-1"
            ],
            [
                "B2RX88-1",
                "A0A087WRQ0-1",
                "1",
                "1",
                "tr|A0A087WRQ0-1|A0A087WRQ0_MOUSE Computationally mapped isoform of Centrosome and spindle pole-associated protein 1  OS=Mus musculus OX=10090 GN=Cspp1 PE=1 SV=1 CANONICAL=B2RX88-1"
            ],
            [
                "A0A087WRQ3-1",
                "A0A075B5Z4-1",
                "4",
                "1",
                "tr|A0A075B5Z4-1|A0A087WRQ3_MOUSE Computationally mapped isoform of T cell receptor alpha variable 1  OS=Mus musculus OX=10090 GN=Trav1 PE=4 SV=1 CANONICAL=A0A087WRQ3-1"
            ],
            [
                "A0A087WRQ3-1",
                "A0A087WRQ3-1",
                "4",
                "1",
                "tr|A0A087WRQ3-1|A0A087WRQ3_MOUSE Isoform 1 of T cell receptor alpha variable 1  OS=Mus musculus OX=10090 GN=Trav1 PE=4 SV=1 CANONICAL=A0A087WRQ3-1"
            ],
            [
                "Q9CXG9-1",
                "A0A0J9YU15-1",
                "1",
                "1",
                "tr|A0A0J9YU15-1|PHF19_MOUSE Computationally mapped isoform of PHD finger protein 19  OS=Mus musculus OX=10090 GN=Phf19 PE=1 SV=1 CANONICAL=Q9CXG9-1"
            ],
            [
                "Q9CXG9-1",
                "Q9CXG9-1",
                "1",
                "1",
                "sp|Q9CXG9-1|PHF19_MOUSE Isoform 1 of PHD finger protein 19  OS=Mus musculus OX=10090 GN=Phf19 PE=1 SV=1 CANONICAL=Q9CXG9-1"
            ],
            [
                "Q07968-1",
                "Q07968-1",
                "1",
                "2",
                "sp|Q07968-1|F13B_MOUSE Isoform 1 of Coagulation factor XIII B chain  OS=Mus musculus OX=10090 GN=F13b PE=1 SV=2 CANONICAL=Q07968-1"
            ],
            [
                "A5H0M4-1",
                "A5H0M4-1",
                "2",
                "1",
                "tr|A5H0M4-1|A5H0M4_MOUSE Isoform 1 of POTE ankyrin domain family, member G  OS=Mus musculus OX=10090 GN=Poteg PE=2 SV=1 CANONICAL=A5H0M4-1"
            ],
            [
                "Q9CXG3-1",
                "D6RJ77-1",
                "1",
                "1",
                "tr|D6RJ77-1|PPIL4_MOUSE Computationally mapped isoform of Peptidyl-prolyl cis-trans isomerase-like 4  OS=Mus musculus OX=10090 GN=Ppil4 PE=1 SV=1 CANONICAL=Q9CXG3-1"
            ],
            [
                "Q9CXG3-1",
                "Q9CXG3-1",
                "1",
                "2",
                "sp|Q9CXG3-1|PPIL4_MOUSE Isoform 1 of Peptidyl-prolyl cis-trans isomerase-like 4  OS=Mus musculus OX=10090 GN=Ppil4 PE=1 SV=2 CANONICAL=Q9CXG3-1"
            ],
            [
                "A6PWV5-1",
                "B7ZP08-1",
                "2",
                "1",
                "tr|B7ZP08-1|B7ZP08_MOUSE Computationally mapped isoform of Arid3c protein  OS=Mus musculus OX=10090 GN=Arid3c PE=2 SV=1 CANONICAL=A6PWV5-1"
            ],
            [
                "B1AQ89-1",
                "B1AQ89-1",
                "4",
                "1",
                "tr|B1AQ89-1|B1AQ89_MOUSE Isoform 1 of Predicted gene 11555  OS=Mus musculus OX=10090 GN=Gm11555 PE=4 SV=1 CANONICAL=B1AQ89-1"
            ],
            [
                "B1AQ89-1",
                "E9PUW3-1",
                "4",
                "1",
                "tr|E9PUW3-1|B1AQ89_MOUSE Computationally mapped isoform of Predicted gene 11555  OS=Mus musculus OX=10090 GN=Gm11555 PE=4 SV=1 CANONICAL=B1AQ89-1"
            ],
            [
                "Q8R5F8-1",
                "F7BBD2-1",
                "4",
                "1",
                "tr|F7BBD2-1|F7BBD2_MOUSE Computationally mapped isoform of Epidermal growth factor receptor kinase substrate 8-like protein 1  OS=Mus musculus OX=10090 GN=Eps8l1 PE=4 SV=1 CANONICAL=Q8R5F8-1"
            ],
            [
                "Q9R099-1",
                "F8WI46-1",
                "1",
                "1",
                "tr|F8WI46-1|F8WI46_MOUSE Computationally mapped isoform of Transducin beta-like protein 2  OS=Mus musculus OX=10090 GN=Tbl2 PE=1 SV=1 CANONICAL=Q9R099-1"
            ],
            [
                "A0A1B0GST8-1",
                "A0A1B0GST8-1",
                "1",
                "1",
                "tr|A0A1B0GST8-1|A0A1B0GST8_MOUSE Isoform 1 of Uncharacterized protein  OS=Mus musculus OX=10090 GN= PE=1 SV=1 CANONICAL=A0A1B0GST8-1"
            ],
            [
                "Q6P549-1",
                "A0A1B0GST7-1",
                "1",
                "1",
                "tr|A0A1B0GST7-1|A0A1B0GST7_MOUSE Computationally mapped isoform of Phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 2  OS=Mus musculus OX=10090 GN=Inppl1 PE=1 SV=1 CANONICAL=Q6P549-1"
            ],
            [
                "Q61470-1",
                "A0A1B0GST6-1",
                "1",
                "1",
                "tr|A0A1B0GST6-1|A0A1B0GST6_MOUSE Computationally mapped isoform of Leukocyte antigen CD37  OS=Mus musculus OX=10090 GN=Cd37 PE=1 SV=1 CANONICAL=Q61470-1"
            ],
            [
                "O35400-1",
                "A0A1B0GST5-1",
                "1",
                "1",
                "tr|A0A1B0GST5-1|A0A1B0GST5_MOUSE Computationally mapped isoform of Sulfotransferase  OS=Mus musculus OX=10090 GN=Sult2b1 PE=1 SV=1 CANONICAL=O35400-1"
            ],
            [
                "B1AQ87-1",
                "B1AQ87-1",
                "4",
                "1",
                "tr|B1AQ87-1|B1AQ87_MOUSE Isoform 1 of Predicted gene 14190  OS=Mus musculus OX=10090 GN=Gm14190 PE=4 SV=1 CANONICAL=B1AQ87-1"
            ],
            [
                "B1AQ85-1",
                "B1AQ85-1",
                "4",
                "1",
                "tr|B1AQ85-1|B1AQ85_MOUSE Isoform 1 of Keratin-associated protein 4-2  OS=Mus musculus OX=10090 GN=Krtap4-2 PE=4 SV=1 CANONICAL=B1AQ85-1"
            ],
            [
                "A0A1B0GST0-1",
                "A0A1B0GST0-1",
                "3",
                "1",
                "tr|A0A1B0GST0-1|A0A1B0GST0_MOUSE Isoform 1 of Olfactory receptor  OS=Mus musculus OX=10090 GN=Olfr566 PE=3 SV=1 CANONICAL=A0A1B0GST0-1"
            ],
            [
                "A0A1B0GST0-1",
                "Q7TRS5-1",
                "3",
                "1",
                "tr|Q7TRS5-1|A0A1B0GST0_MOUSE Computationally mapped isoform of Olfactory receptor  OS=Mus musculus OX=10090 GN=Olfr566 PE=3 SV=1 CANONICAL=A0A1B0GST0-1"
            ],
            [
                "Q3UU21-1",
                "Q3UU21-1",
                "2",
                "1",
                "tr|Q3UU21-1|Q3UU21_MOUSE Isoform 1 of Zona pellucida-like domain-containing 2  OS=Mus musculus OX=10090 GN=Zpld2 PE=2 SV=1 CANONICAL=Q3UU21-1"
            ],
            [
                "P01942-1",
                "P01942-1",
                "1",
                "2",
                "sp|P01942-1|HBA_MOUSE Isoform 1 of Hemoglobin subunit alpha  OS=Mus musculus OX=10090 GN=Hba PE=1 SV=2 CANONICAL=P01942-1"
            ],
            [
                "P01942-1",
                "Q91VB8-1",
                "1",
                "1",
                "tr|Q91VB8-1|HBA_MOUSE Computationally mapped isoform of Hemoglobin subunit alpha  OS=Mus musculus OX=10090 GN=Hba PE=1 SV=1 CANONICAL=P01942-1"
            ],
            [
                "P03899-1",
                "P03899-1",
                "1",
                "3",
                "sp|P03899-1|NU3M_MOUSE Isoform 1 of NADH-ubiquinone oxidoreductase chain 3  OS=Mus musculus OX=10090 GN=Mtnd3 PE=1 SV=3 CANONICAL=P03899-1"
            ],
            [
                "Q3TC93-1",
                "A0A1W2P6R8-1",
                "1",
                "1",
                "tr|A0A1W2P6R8-1|A0A1W2P6R8_MOUSE Computationally mapped isoform of HCLS1-binding protein 3  OS=Mus musculus OX=10090 GN=Hs1bp3 PE=1 SV=1 CANONICAL=Q3TC93-1"
            ],
            [
                "Q9CWR0-1",
                "A0A1W2P6R9-1",
                "1",
                "1",
                "tr|A0A1W2P6R9-1|A0A1W2P6R9_MOUSE Computationally mapped isoform of Rho guanine nucleotide exchange factor 25  OS=Mus musculus OX=10090 GN=Arhgef25 PE=1 SV=1 CANONICAL=Q9CWR0-1"
            ],
            [
                "O35153-1",
                "A0A1B0GSF3-1",
                "3",
                "1",
                "tr|A0A1B0GSF3-1|BET1L_MOUSE Computationally mapped isoform of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "O35153-1",
                "A0A1B0GR52-1",
                "3",
                "1",
                "tr|A0A1B0GR52-1|BET1L_MOUSE Computationally mapped isoform of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "O35153-1",
                "O35153-1",
                "3",
                "1",
                "sp|O35153-1|BET1L_MOUSE Isoform 1 of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "O35153-1",
                "A0A1B0GQV3-1",
                "3",
                "1",
                "tr|A0A1B0GQV3-1|BET1L_MOUSE Computationally mapped isoform of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "O35153-1",
                "A0A1B0GR53-1",
                "3",
                "1",
                "tr|A0A1B0GR53-1|BET1L_MOUSE Computationally mapped isoform of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "O35153-1",
                "A0A1B0GS34-1",
                "3",
                "1",
                "tr|A0A1B0GS34-1|BET1L_MOUSE Computationally mapped isoform of BET1-like protein  OS=Mus musculus OX=10090 GN=Bet1l PE=3 SV=1 CANONICAL=O35153-1"
            ],
            [
                "A2RSJ4-1",
                "A0A1W2P6R4-1",
                "4",
                "1",
                "tr|A0A1W2P6R4-1|A0A1W2P6R4_MOUSE Computationally mapped isoform of UHRF1-binding protein 1-like  OS=Mus musculus OX=10090 GN=Uhrf1bp1l PE=4 SV=1 CANONICAL=A2RSJ4-1"
            ],
            [
                "Q91YE9-1",
                "A0A1W2P6R5-1",
                "4",
                "1",
                "tr|A0A1W2P6R5-1|A0A1W2P6R5_MOUSE Computationally mapped isoform of Cytosolic 5'-nucleotidase 1B  OS=Mus musculus OX=10090 GN=Nt5c1b PE=4 SV=1 CANONICAL=Q91YE9-1"
            ],
            [
                "Q02853-1",
                "A0A1W2P6R6-1",
                "4",
                "1",
                "tr|A0A1W2P6R6-1|A0A1W2P6R6_MOUSE Computationally mapped isoform of Stromelysin-3  OS=Mus musculus OX=10090 GN=Mmp11 PE=4 SV=1 CANONICAL=Q02853-1"
            ],
            [
                "Q8BRF7-1",
                "A0A1W2P6R7-1",
                "1",
                "1",
                "tr|A0A1W2P6R7-1|A0A1W2P6R7_MOUSE Computationally mapped isoform of Sec1 family domain-containing protein 1  OS=Mus musculus OX=10090 GN=Scfd1 PE=1 SV=1 CANONICAL=Q8BRF7-1"
            ],
            [
                "Q6DYE8-1",
                "A0A1W2P6R0-1",
                "1",
                "1",
                "tr|A0A1W2P6R0-1|A0A1W2P6R0_MOUSE Computationally mapped isoform of Ectonucleotide pyrophosphatase/phosphodiesterase family member 3  OS=Mus musculus OX=10090 GN=Enpp3 PE=1 SV=1 CANONICAL=Q6DYE8-1"
            ],
            [
                "L0N7N1-1",
                "L0N759-1",
                "1",
                "1",
                "tr|L0N759-1|L0N759_MOUSE Computationally mapped isoform of Kinesin-like protein KIF14  OS=Mus musculus OX=10090 GN=Kif14 PE=1 SV=1 CANONICAL=L0N7N1-1"
            ],
            [
                "Q9R092-1",
                "A0A1W2P6R2-1",
                "4",
                "1",
                "tr|A0A1W2P6R2-1|A0A1W2P6R2_MOUSE Computationally mapped isoform of 17-beta-hydroxysteroid dehydrogenase type 6  OS=Mus musculus OX=10090 GN=Hsd17b6 PE=4 SV=1 CANONICAL=Q9R092-1"
            ],
            [
                "Q9D7Y9-1",
                "A2AHF0-1",
                "4",
                "1",
                "tr|A2AHF0-1|A2AHF0_MOUSE Computationally mapped isoform of Protein SLX4IP  OS=Mus musculus OX=10090 GN=Slx4ip PE=4 SV=1 CANONICAL=Q9D7Y9-1"
            ],
            [
                "Q9QYH9-1",
                "Q9QYH9-1",
                "2",
                "1",
                "sp|Q9QYH9-1|TNF14_MOUSE Isoform 1 of Tumor necrosis factor ligand superfamily member 14  OS=Mus musculus OX=10090 GN=Tnfsf14 PE=2 SV=1 CANONICAL=Q9QYH9-1"
            ],
            [
                "P52479-1",
                "P52479-1",
                "1",
                "3",
                "sp|P52479-1|UBP10_MOUSE Isoform 1 of Ubiquitin carboxyl-terminal hydrolase 10  OS=Mus musculus OX=10090 GN=Usp10 PE=1 SV=3 CANONICAL=P52479-1"
            ],
            [
                "P52479-1",
                "P52479-2",
                "1",
                "3",
                "sp|P52479-2|UBP10_MOUSE Isoform 2 of Ubiquitin carboxyl-terminal hydrolase 10  OS=Mus musculus OX=10090 GN=Usp10 PE=1 SV=3 CANONICAL=P52479-1"
            ],
            [
                "P52479-1",
                "G3X9U6-1",
                "1",
                "1",
                "tr|G3X9U6-1|UBP10_MOUSE Computationally mapped isoform of Ubiquitin carboxyl-terminal hydrolase 10  OS=Mus musculus OX=10090 GN=Usp10 PE=1 SV=1 CANONICAL=P52479-1"
            ],
            [
                "P52479-1",
                "G3UYQ8-1",
                "1",
                "1",
                "tr|G3UYQ8-1|UBP10_MOUSE Computationally mapped isoform of Ubiquitin carboxyl-terminal hydrolase 10  OS=Mus musculus OX=10090 GN=Usp10 PE=1 SV=1 CANONICAL=P52479-1"
            ],
            [
                "A0A0U1RNG7-1",
                "F7DE04-1",
                "1",
                "1",
                "tr|F7DE04-1|F7DE04_MOUSE Computationally mapped isoform of Potassium channel tetramerisation domain-containing 14  OS=Mus musculus OX=10090 GN=Kctd14 PE=1 SV=1 CANONICAL=A0A0U1RNG7-1"
            ],
            [
                "B1AXU1-1",
                "B1AXU1-1",
                "4",
                "1",
                "tr|B1AXU1-1|B1AXU1_MOUSE Isoform 1 of Predicted gene 6812  OS=Mus musculus OX=10090 GN=Gm6812 PE=4 SV=1 CANONICAL=B1AXU1-1"
            ],
            [
                "P02762-1",
                "P02762-1",
                "1",
                "2",
                "sp|P02762-1|MUP6_MOUSE Isoform 1 of Major urinary protein 6  OS=Mus musculus OX=10090 GN=Mup6 PE=1 SV=2 CANONICAL=P02762-1"
            ],
            [
                "P02762-1",
                "A9C497-1",
                "1",
                "1",
                "tr|A9C497-1|MUP6_MOUSE Computationally mapped isoform of Major urinary protein 6  OS=Mus musculus OX=10090 GN=Mup6 PE=1 SV=1 CANONICAL=P02762-1"
            ],
            [
                "Q9QYH6-1",
                "Q9QYH6-1",
                "1",
                "1",
                "sp|Q9QYH6-1|MAGD1_MOUSE Isoform 1 of Melanoma-associated antigen D1  OS=Mus musculus OX=10090 GN=Maged1 PE=1 SV=1 CANONICAL=Q9QYH6-1"
            ]
        ]
    },
    "readme": "Name\n   - Mouse Protein Sequence Info (UniProtKB)\n\nVersion\n   - 1.11.2\n\nCreated on\n  - 2020-03-15T14:34:41.443981\n\nModified on\n   - 2020-03-15T15:16:13.169046\n\nCreated by\n   - Jeet Vora [jeetvora@gwu.edu], Rahi Navelkar [rsn13@gwu.edu], Robel Kahsay [rykahsay@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   - Maria Martin [martin@ebi.ac.uk], Preethi Vasudev [pvasudev@ebi.ac.uk]\n\nChecksum\n   - \n\nUsability Domain\n   - The Mouse Protein Sequence Info (UniProtKB) dataset contains mouse [taxid:10090] UniProtKB protein fasta sequence information that includes sequence version and fasta header. 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 sequence info\n   - fasta header\n   - UniProtKB\n\n   Pipeline Steps\n   - Step-1: The script loads the UniProtKB sequence for all mouse isoform accessions data from the triplestore containing triples from nt file and extracts it into a dataset in a fasta format\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/ln2wwwdata/source/uniprot-proteome-mus-musculus.nt\n   Output Subdomain\n      Name: mouse_protein_sequenceinfo.csv\n      URI: http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_sequenceinfo.csv\n      Name: mouse_protein_sequenceinfo.stat.csv\n      URI: http://data.glygen.org/ln2wwwdata/reviewed/mouse_protein_mouse_sequenceinfo.stat.csv\n\nFields\n      Name: \"uniprotkb_canonical_ac\"\n     Description: \"\"\n      Unique Values: \"16\"\n      Name: \"motif_start\"\n     Description: \"\"\n      Unique Values: \"23\"\n      Name: \"filter_flags\"\n     Description: \"Filter tags/flags assigned to the mutation entries that have passed the mutation filter criteria\"\n      Unique Values: \"19\"\n      Name: \"sift_prediction\"\n     Description: \"The SIFT qualitative prediction is derived from this score such that substitutions with a score < 0.05 are called 'deleterious' and all others are called 'tolerated\"\n      Unique Values: \"3\"\n      Name: \"somatic_status\"\n     Description: \"The status of the variant whether it is germline (somatic status =0) or somatic (somatic status =1) variant\"\n      Unique Values: \"1\"\n      Name: \"chr_pos\"\n     Description: \"Chromosome Position of a gene\"\n      Unique Values: \"19\"\n      Name: \"chr_id\"\n     Description: \"\"\n      Unique Values: \"10\"\n      Name: \"dbsnp_id\"\n     Description: \"\"A reference SNP ID number\"\n      Unique Values: \"19\"\n      Name: \"evidence_ECO:0000313\"\n     Description: \"A type of evidence that is imported information used in an automatic assertion\"\n      Unique Values: \"1\"\n      Name: \"xref_id\"\n     Description: \"ID of the X-ref\"\n      Unique Values: \"19\"\n      Name: \"mim_id\"\n     Description: \"Disease ID from the OMIM database\"\n      Unique Values: \"1\"\n      Name: \"description\"\n     Description: \"\"\n      Unique Values: \"1\"\n      Name: \"begin_aa_pos\"\n     Description: \"Start position of the amino acid in the sequence of a protein in UniProtKB\"\n      Unique Values: \"18\"\n      Name: \"cytogenic_band\"\n     Description: \"One of the subregions of a chromosome visible microscopically after special staining\"\n      Unique Values: \"14\"\n      Name: \"aa_pos\"\n     Description: \"Position of the amino acid in canonical sequence of protein in the UniProtKB database\"\n      Unique Values: \"18\"\n      Name: \"ref_aa\"\n     Description: \"Reference amino acid residue\"\n      Unique Values: \"9\"\n      Name: \"sift_score\"\n     Description: \"SIFT predicts whether an amino acid substitution is likely to affect protein function based on sequence homology and the physico-chemical similarity between the alternate amino acids. The data we provide for each amino acid substitution is a score and a qualitative prediction (either 'tolerated' or 'deleterious'). The score is the normalized probability that the amino acid change is tolerated so scores nearer zero are more likely to be deleterious. The qualitative prediction is derived from this score such that substitutions with a score < 0.05 are called 'deleterious' and all others are called 'tolerated\"\n      Unique Values: \"10\"\n      Name: \"alt_motif\"\n     Description: \"Altered motif\"\n      Unique Values: \"24\"\n      Name: \"disease_xrefs\"\n     Description: \"\"\n      Unique Values: \"1\"\n      Name: \"polyphen_score\"\n     Description: \"The score assigned to a amino acid substitution by a tool called as PolyPhen (=Polymorphism Phenotyping)  which predicts possible impact of an amino acid substitution on the structure and function of a human protein using straightforward physical and comparative considerations. The PolyPhen score ranges from 0.0 (tolerated) to 1.0 (deleterious). Variants with scores of 0.0 are predicted to be benign. Values closer to 1.0 are more confidently predicted to be deleterious\"\n      Unique Values: \"1\"\n      Name: \"polyphen_prediction\"\n     Description: \"\"The PolyPhen prediction values are - greater than 0.908 - \"\"Probably Damaging\"\"; greater than 0.446 and less than or equal to 0.908 - \"\"Possibly Damaging\"\"; less than or equal to 0.446 - \"\"Benign\"\"\"\n      Unique Values: \"1\"\n      Name: \"is_known_glycoprotein\"\n     Description: \"Yes or no flag to denote if a protein is experimentally proven glycoprotein\"\n      Unique Values: \"1\"\n      Name: \"alt_nt\"\n     Description: \"Alternative nucleotide\"\n      Unique Values: \"1\"\n      Name: \"protein_name\"\n     Description: \"Protein name in the UniProtKB\"\n      Unique Values: \"16\"\n      Name: \"disease_evidence_ECO:0000313\"\n     Description: \"The evidence that could be a PMID or a ClinVar ID that associates disease with the variant\"\n      Unique Values: \"1\"\n      Name: \"effect\"\n     Description: \"The effect on glycosylation or N-sequon due to variation at site of glycosylated amino acid.\"\n      Unique Values: \"3\"\n      Name: \"end_aa_pos\"\n     Description: \"End position of the amino acid in the sequence of a protein in UniProtKB\"\n      Unique Values: \"18\"\n      Name: \"mutation_type\"\n     Description: \"Denotes whether mutation was reported with reference to amino acid or nucleotide\"\n      Unique Values: \"1\"\n      Name: \"disease_description\"\n     Description: \"The disease description and/or definition \"\n      Unique Values: \"1\"\n      Name: \"do_id\"\n     Description: \"Disease Ontology identifier\"\n      Unique Values: \"1\"\n      Name: \"evidence_ECO:0000269\"\n     Description: \"A type of experimental evidence that is used in a manual assertion. Usually the values are PMIDs\"\n      Unique Values: \"1\"\n      Name: \"ref_nt\"\n     Description: \"Reference nucleotide\"\n      Unique Values: \"4\"\n      Name: \"disease\"\n     Description: \"The disease name for the cell line\"\n      Unique Values: \"1\"\n      Name: \"alt_aa\"\n     Description: \"Altered amino acid residue\"\n      Unique Values: \"10\"\n      Name: \"cosmic_id\"\n     Description: \"Legacy mutation identifier (COSM) in Catalogue of Somatic Mutations in Cancer (COSMIC) represents existing COSM mutation identifiers. This identifier remains the same between different assemblies (GRCh37 and GRCh38). All the COSM ids at the same genomic location have been collapsed into one representative COSM id. These ids are maintained to help track existing mutations\"\n      Unique Values: \"1\"\n      Name: \"ref_motif\"\n     Description: \"Reference motif\"\n      Unique Values: \"20\"\n      Name: \"xref_key\"\n     Description: \"Source key associated to the  xref_id\"\n      Unique Values: \"2\"\n      Name: \"minor_allelic_frequency\"\n     Description: \"\"\n      Unique Values: \"1\"\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: \"17\"\n      Name: \"disease_evidence_ECO:0000269\"\n     Description: \"The PMID evidence for the association of the disease with the variant.\"\n      Unique Values: \"1\"",
    "downloadurl": "/ln2data/releases/data/v-1.11.2/reviewed/mouse_protein_sequenceinfo.csv"
}