profiles.sql

CREATE TABLE antismash.profiles (
    name	text NOT NULL,
    description	text,
    cutoff	int4,
    filename	text,
    CONSTRAINT profiles_pkey PRIMARY KEY (name)
);

COMMENT ON TABLE antismash.profiles IS
  'HMM profiles used during cluster detection';

INSERT INTO antismash.profiles (name, description, cutoff, filename)
VALUES
    ('Condensation', 'Condensation domain', 20, 'Condensation.hmm'),
    ('AMP-binding', 'Adenylation domain', 20, 'AMP-binding.hmm'),
    ('A-OX', 'Adenylation domain with integrated oxidase', 500, 'A-OX.hmm'),
    ('PP-binding', 'Thiolation domain', 20, 'PP-binding.hmm'),
    ('PKS_KS', 'Ketosynthase domain', 50, 'PKS_KS.hmm'),
    ('PKS_AT', 'Acyltransferase domain', 50, 'PKS_AT.hmm'),
    ('ATd', 'Trans-acyltransferase docking domain', 65, 'ATd.hmm'),
    ('bt1fas', 'Bacterial type I fatty acid synthase', 50, 'bt1fas.hmm'),
    ('ft1fas', 'Fungal type I fatty acid synthase', 50, 'ft1fas.hmm'),
    ('t2fas', 'Type II fatty acid synthase', 50, 't2fas.hmm'),
    ('fabH', 'FabH fatty acid synthase', 50, 'fabH.hmm'),
    ('ene_KS', 'Enediyine ketosynthase', 50, 'ene_KS.hmm'),
    ('mod_KS', 'Modular ketosynthase', 50, 'mod_KS.hmm'),
    ('hyb_KS', 'Hybrid ketosynthase', 50, 'hyb_KS.hmm'),
    ('itr_KS', 'Iterative ketosynthase', 50, 'itr_KS.hmm'),
    ('tra_KS', 'Trans-AT ketosynthase', 50, 'tra_KS.hmm'),
    ('hglD', 'Unusual PKS HglD-like', 50, 'hglD.hmm'),
    ('hglE', 'Unusual PKS HglE-like', 50, 'hglE.hmm'),
    ('t2ks', 'Type II PKS ketosynthase', 50, 't2ks.hmm'),
    ('t2pks2', 'Type II PKS ketosynthase, model 2', 450, 't2ks2.hmm'),
    ('t2clf', 'Type II PKS Chain length factor', 50, 't2clf.hmm'),
    ('Chal_sti_synt_N', 'Type III PKS N-terminal', 21, 'Chal_sti_synt_N.hmm'),
    ('Chal_sti_synt_C', 'Type III PKS C-terminal', 21, 'Chal_sti_synt_C.hmm'),
    ('phytoene_synt', 'Phytoene_synthase', 20, 'phytoene_synt.hmm'),
    ('Lycopene_cycl', 'Lycopene cyclase', 80, 'Lycopene_cycl.hmm'),
    ('dmat', 'Dimethylallyl tryptophan synthase', 200, 'dmat.hmm'),
    ('LANC_like', 'LanC-like lantibiotics biosynthesis protein', 17, 'LANC_like.hmm'),
    ('DUF4135', 'Lantibiotic-associated domain', 150, 'PF13575.hmm'),
    ('Lant_dehydr_N', 'Lantibiotic dehydratase, N-terminus', 27, 'PF04738.hmm'),
    ('Lant_dehydr_C', 'Lantibiotic dehydratase, C-terminus', 27, 'PF14028.hmm'),
    ('Flavoprotein', 'Lantibiotic aminovinly flavoprotein', 20, 'PF02441.hmm'),
    ('Trp_halogenase', 'Tryptophan halogenase', 20, 'PF04820.hmm'),
    ('p450', 'P450 oxygenase', 21, 'PF00067.hmm'),
    ('Pkinase', 'Protein kinase domain', 30, 'PF00069.hmm'),
    ('adh_short', 'Short-chain dehydrogenase', 100, 'PF00106.hmm'),
    ('adh_short_C2', 'Short-chain dehydrogenase, C-terminus', 100, 'PF13561.hmm'),
    ('PCMT', 'Protein-L-isoaspartate(D-aspartate) O-methyltransferase (PCMT)', 20, 'PF01135.hmm'),
    ('Antimicr18', 'Lantibiotic antimicrobial peptide 18', 20, 'Antimicrobial18.hmm'),
    ('Gallidermin', 'Gallidermin', 20, 'Gallidermin.hmm'),
    ('L_biotic_A', 'Lantibiotic, type A', 20, 'L_biotic_typeA.hmm'),
    ('TIGR03731', 'Lantibiotic, gallidermin/nisin family', 20, 'TIGR03731.hmm'),
    ('leader_d', 'Lantibiotic leader lacticin 481 group', 20, 'LE-LAC481.hmm'),
    ('leader_eh', 'Lantibiotic leader mersacidin cinnamycin group', 20, 'LE-MER+2PEP.hmm'),
    ('leader_abc', 'Lantibiotic leader LanBC modified', 20, 'LE-LanBC.hmm'),
    ('mature_d', 'Lantibiotic peptide lacticin 481 group', 20, 'MA-LAC481.hmm'),
    ('mature_ab', 'Lantibiotic peptide nisin epidermin group', 20, 'MA-NIS+EPI.hmm'),
    ('mature_a', 'Lantibiotic peptide nisin group', 20, 'MA-NIS.hmm'),
    ('mature_b', 'Lantibiotic peptide epidermin group', 25, 'MA-EPI.hmm'),
    ('mature_ha', 'Lantibiotic peptide two component alpha', 20, 'MA-2PEPA.hmm'),
    ('mature_h_beta', 'Lantibiotic peptide two component beta', 20, 'MA-2PEPB.hmm'),
    ('lacticin_l', 'lantibiotic peptide lacticin 481 group (dufour et al)', 20, 'LE-DUF.hmm'),
    ('lacticin_mat', 'lantibiotic leader lacticin 481 group (dufour et al)', 20, 'MA-DUF.hmm'),
    ('LD_lanti_pre', 'FxLD family lantipeptide', 20, 'TIGR04363.hmm'),
    ('strep_PEQAXS', 'Streptomyces PEQAXS motif lantipeptide', 20, 'strep_PEQAXS.hmm'),
    ('strepbact', 'Putative Streptomyces bacteriocin', 50, 'strepbact.hmm'),
    ('Antimicrobial14', 'Antimicrobial peptide 14', 90, 'Antimicrobial14.hmm'),
    ('Bacteriocin_IIc', 'Bacteriocin_IIc domain', 27, 'Bacteriocin_IIc.hmm'),
    ('Bacteriocin_IId', 'Bacteriocin_IId domain', 23, 'Bacteriocin_IId.hmm'),
    ('BacteriocIIc_cy', 'Bacteriocin_IIdc_cydomain', 92, 'BacteriocIIc_cy.hmm'),
    ('Bacteriocin_II', 'Bacteriocin_II domain', 40, 'Bacteriocin_II.hmm'),
    ('Bacteriocin_IIi', 'Bacteriocin_IIi domain', 56, 'Bacteriocin_IIi.hmm'),
    ('Lactococcin', 'Lactococcin', 24, 'Lactococcin.hmm'),
    ('Antimicrobial17', 'Antimicrobial peptide 17', 31, 'Antimicrobial17.hmm'),
    ('Lactococcin_972', 'Lactococcin 972', 25, 'Lactococcin_972.hmm'),
    ('LcnG-beta', 'Lactococcin G-beta', 78, 'LcnG-beta.hmm'),
    ('Subtilosin_A', 'Subtilosin A', 68, 'Subtilosin_A.hmm'),
    ('Cloacin', 'Cloacin', 27, 'Cloacin.hmm'),
    ('Linocin_M18', 'Linocin M18', 25, 'Linocin_M18.hmm'),
    ('TIGR03651', 'Bacteriocin, circularin A/uberolysin famil', 18, 'TIGR03651.hmm'),
    ('TIGR03678', 'Bacteriocin, microcyclamide/patellamide family', 35, 'TIGR03678.hmm'),
    ('TIGR03693', 'Thiazole-containing bacteriocin maturation protein', 400, 'TIGR03693.hmm'),
    ('TIGR03798', 'Bacteriocin propeptide', 20, 'TIGR03798.hmm'),
    ('TIGR03882', 'Bacteriocin biosynthesis cyclodehydratase', 150, 'TIGR03882.hmm'),
    ('TIGR03601', 'Bacteriocin, BA_2677 family', 50, 'TIGR03601.hmm'),
    ('TIGR03795', 'Ribosomal natural product, two-chain TOMM family', 41, 'TIGR03795.hmm'),
    ('TIGR03793', 'NHLP leader peptide domain', 51, 'TIGR03793.hmm'),
    ('TIGR03975', 'Bacteriocin maturation radical SAM protein 1', 282, 'TIGR03975.hmm'),
    ('TIGR01193', 'ABC-type bacteriocin transporter', 523, 'TIGR01193.hmm'),
    ('mvnA', 'Microviridin A', 25, 'mvnA.hmm'),
    ('thiostrepton', 'Thiostrepton-like thiopeptides', 20, 'thiostrepton.hmm'),
    ('atropopeptide_p450', 'atropopeptide-specific cytochrome p450', 500, 'atropopeptide_p450.hmm'),
    ('BLS', 'Beta-lactam synthase', 250, 'BLS.hmm'),
    ('CAS', 'Clavulanic acid synthase-like', 250, 'CAS.hmm'),
    ('tabtoxin', 'Tabtoxin synthase-like', 500, 'tabtoxin.hmm'),
    ('DOIS', '2-deoxy-scyllo-inosose synthase', 500, 'DOIS.hmm'),
    ('NeoL', 'NeoL-like deacetylase', 50, 'neoL_like.hmm'),
    ('SpcDK_glyc', 'SpcD-/SpcK-like thymidylyltransferas', 600, 'spcDK_like_glyc.hmm'),
    ('SpcFG', 'SpcF-/SpcG-like glycosyltransferase', 200, 'spcFG_like.hmm'),
    ('strH', 'StrH-like glycosyltransferase', 200, 'strH_like.hmm'),
    ('strK1', 'StrK-like phosphatase', 800, 'strK_like1.hmm'),
    ('strK2', 'StrK-like phosphatase', 650, 'strK_like2.hmm'),
    ('ValA', 'ValA-like 2-epi-5-epi-valiolone synthase', 600, 'valA_like.hmm'),
    ('salQ', '2-epi-5-epi-valiolone synthase, SalQ-like', 480, 'salQ.hmm'),
    ('novK', 'NovK-like reductase', 200, 'novK.hmm'),
    ('novJ', 'NovJ-like reductase', 350, 'novJ.hmm'),
    ('novI', 'NovI-like cytochrome P450', 600, 'novI.hmm'),
    ('novH', 'NovH-like protein', 750, 'novH.hmm'),
    ('SpcDK_cou', 'SpcD/SpcK-like thymidylyltransferase, aminocoumarins group', 600, 'spcDK_like_cou.hmm'),
    ('IucA_IucC', 'Siderophore synthase', 30, 'IucA_IucC.hmm'),
    ('ectoine_synt', 'Ectoine synthase', 35, 'ectoine_synt.hmm'),
    ('AfsA', 'AfsA-like butyrolactone synthase', 25, 'AfsA.hmm'),
    ('indsynth', 'StaD-like chromopyrrolic acid synthase domain', 100, 'indsynth.hmm'),
    ('LipM', 'LipM-like nucleotidyltransferase', 50, 'LipM.hmm'),
    ('LipU', 'LipU-like protein', 30, 'LipU.hmm'),
    ('LipV', 'LipV-like dehydrogenase', 375, 'LipV.hmm'),
    ('TunD', 'TunD-like putative N-acetylglucosamine transferase', 200, 'TunD.hmm'),
    ('pur6', 'Pur6-like synthetase', 200, 'pur6.hmm'),
    ('pur10', 'Pur10-like oxidoreductase', 600, 'pur10.hmm'),
    ('nikJ', 'NikJ-like protein', 200, 'nikJ.hmm'),
    ('nikO', 'NikO-like enolpyruvyl transferase', 400, 'nikO.hmm'),
    ('MoeO5', 'MoeO5-like prenyl-3-phosphoglycerate synthase', 65, 'MoeO5.hmm'),
    ('moeGT', 'Phosphoglycolipid glycosyltransferase', 100, 'moeGT.hmm'),
    ('melC', 'MelC-like melanin synthase', 40, 'melC.hmm'),
    ('NAD_binding_4', 'NAD-binding domain 4', 40, 'NAD_binding_4.hmm'),
    ('LmbU', 'LmbU-like protein', 50, 'LmbU.hmm'),
    ('mmyO', 'MmyO-like protein', 500, 'mmyO.hmm'),
    ('AvrD', 'Pseudomonas avriulence D protein', 50, 'PF05655.hmm'),
    ('goadsporin_like', 'Goadsporin-like protein', 500, 'goadsporin_like.hmm'),
    ('LeupA', 'leupeptin A', 580, 'LeupA.hmm'),
    ('LeupB', 'leupeptin B', 800, 'LeupB.hmm'),
    ('LeupC', 'leupeptin C', 450, 'LeupC.hmm'),
    ('LeupD', 'leupeptin D', 225, 'LeupD.hmm'),
    ('TIGR03602', 'Bacteriocin protoxin, streptolysin S family', 50, 'TIGR03602.hmm'),
    ('TIGR03603', 'Bacteriocin biosynthesis cyclodehydratase', 150, 'TIGR03603.hmm'),
    ('TIGR03604', 'Bacteriocin biosynthesis docking scaffold', 219, 'TIGR03604.hmm'),
    ('TIGR03605', 'SagB-type dehydrogenase', 200, 'TIGR03605.hmm'),
    ('Neocarzinostat', 'Neocarzinostatin-like protein', 28, 'Neocarzinostat.hmm'),
    ('cyanobactin_synth', 'Cyanobactin protease', 80, 'cyanobactin_synth.hmm'),
    ('bcpB', 'BcpB-like phosphomutase', 400, 'bcpB.hmm'),
    ('frbD', 'FrbD-like phosphomutase', 350, 'frbD.hmm'),
    ('mitE', 'MitE-like CoA-ligase', 400, 'mitE.hmm'),
    ('vlmB', 'Valanimycin biosynthesis VlmB domain', 250, 'vlmB.hmm'),
    ('prnB', 'Pyrrolnitrin biosynthesis PrnB domain', 200, 'prnB.hmm'),
    ('CaiA', 'Nitrososynthase domain linked to oligosaccharide antibiotics', 200, 'CaiA.hmm'),
    ('MGT', 'Secondary metabolite-related glycosyltransferase', 100, 'MGT.hmm'),
    ('Glycos_transf_1', 'Secondary metabolite-related glycosyltransferase', 20, 'Glycos_transf_1.hmm'),
    ('Glycos_transf_2', 'Secondary metabolite-related glycosyltransferase', 20, 'Glycos_transf_2.hmm'),
    ('Glyco_transf_28', 'Secondary metabolite-related glycosyltransferase', 26, 'Glyco_transf_28.hmm'),
    ('DUF1205', 'Secondary metabolite-related glycosyltransferase', 20, 'DUF1205.hmm'),
    ('Autoind_synth', 'Autoinducer synthetase', 20, 'Autoind_synth.hmm'),
    ('YcaO', 'YcaO-like family', 25, 'YcaO.hmm'),
    ('TIGR01408', 'ubiquitin-activating enzyme E1', 750, 'TIGR01408.hmm'),
    ('PF06968', 'BATS family', 20, 'PF06968.hmm'),
    ('PF04055', 'radical SAM superfamily', 40, 'PF04055.hmm'),
    ('PF07366', 'SnoaL family', 20, 'PF07366.hmm'),
    ('thio_amide', 'nosA tpdK nocA berI pbtH', 20, 'thio_amide.hmm'),
    ('phzB', 'Phenazine biosynthesis gene', 80, 'phzB.hmm'),
    ('phosphonates-like', 'Phosphonate biosynthesis PEP mutase', 275, 'phosphonates-like.hmm'),
    ('phosphonates', 'Phosphonate biosynthesis PEP mutase', 150, 'phosphonates.hmm'),
    ('phosphonatesHydrolase', 'Phosphonopyruvate hydrolase', 400, 'phosphonatesHydrolase.hmm'),
    ('phosphonatesMeIsocitrateLyase', 'Methylisocitrate Lyase', 250, 'phosphonatesMeIsocitrateLyase.hmm'),
    ('2-Hacid_dh_C', 'D-isomer specific 2-hydroxyacid dehydrogenase, NAD binding domain', 25, '2-Hacid_dh_C.hmm'),
    ('Aldedh', 'Aldehyde dehydrogenase family', 23, 'Aldedh.hmm'),
    ('Aminotran_1_2', 'Aminotransferase class I and II', 19, 'Aminotran_1_2.hmm'),
    ('Aminotran_5', 'Aminotransferase class-V', 21, 'Aminotran_5.hmm'),
    ('AurF', 'P-aminobenzoate N-oxygenase AurF', 29, 'AurF.hmm'),
    ('CTP_transf_like', 'Cytidylyltransferase-like', 21, 'CTP_transf_like.hmm'),
    ('DUF4992', 'Domain of unknown function', 29, 'DUF4992.hmm'),
    ('Fe-ADH', 'Iron-containing alcohol dehydrogenase', 26, 'Fe-ADH.hmm'),
    ('Metallophos_2', 'Calcineurin-like phosphoesterase superfamily domain', 27, 'Metallophos_2.hmm'),
    ('NpaA', 'Nitrosuccinic acid synthase', 600, 'NpaA.hmm'),
    ('NpaB', 'Nitrosuccinic acid decarboxylase', 200, 'NpaB.hmm'),
    ('NpaC', 'Nitronate monooxygenase', 400, 'NpaC.hmm'),
    ('NTP_transf_3', 'MobA-like NTP transferase domain', 27, 'NTP_transf_3.hmm'),
    ('NTP_transf_5', 'Uncharacterised nucleotidyltransferase', 31, 'NTP_transf_5.hmm'),
    ('PALP', 'Pyridoxal-phosphate dependent enzyme', 35, 'PALP.hmm'),
    ('PEP_mutase', 'Phosphoenolpyruvate phosphomutase', 31, 'PEP_mutase.hmm'),
    ('TPP_enzyme_C', 'Thiamine pyrophosphate enzyme, C-terminal TPP binding domain', 21, 'TPP_enzyme_C.hmm'),
    ('TPP_enzyme_N', 'Thiamine pyrophosphate enzyme, N-terminal TPP binding domain', 26, 'TPP_enzyme_N.hmm'),
    ('bacilysin', 'Bacilysin-related amino acid ligase', 160, 'bacilysin.hmm'),
    ('subtilosin', 'Putative subtilosin biosynthesis enzyme ywiA', 140, 'subtilosin.hmm'),
    ('sublancin', 'Sublancin prepeptide', 30, 'sublancin.hmm'),
    ('thuricin', 'Thuricin prepeptide', 30, 'thuricin.hmm'),
    ('skfc', 'SkfC biosynthesis enzyme', 70, 'skfc.hmm'),
    ('botH', 'bottromycin biosynthesis enzyme', 65, 'botH.hmm'),
    ('glycocin', 'glycocin prepeptide', 30, 'glycocin.hmm'),
    ('mcjC', 'Microcin J25-processing protein McjC', 60, 'mcjC.hmm'),
    ('micJ25', 'Microcin J25 prepeptide', 60, 'micJ25.hmm'),
    ('DUF692', 'DUF692 domain of unknown function, associated with small prepeptides such as methanobactin', 40, 'DUF692.hmm'),
    ('lasso', 'Lasso peptide modification enzyme', 400, 'lasso.hmm'),
    ('mvd', 'Marinostatin/microviridin prepeptide', 20, 'mvd_mst.hmm'),
    ('cypemycin', 'Cypemycin prepeptide', 20, 'cypemycin.hmm'),
    ('cypI', 'cypemycin biosynthesis cypI enzyme', 20, 'cypI.hmm'),
    ('PoyD', 'proteusin biosynthesis PoyD enzyme', 25, 'PoyD.hmm'),
    ('CBM_48', 'Carbohydrate-binding module 48 (Isoamylase N-terminal domain)', 20, 'PF02922.hmm'),
    ('DegT_DnrJ_EryC1', 'DegT/DnrJ/EryC1/StrS aminotransferase family', 19, 'PF01041.hmm'),
    ('Alpha-amylase', 'Alpha amylase, catalytic domain', 20, 'PF00128.hmm'),
    ('dTDP_sugar_isom', 'dTDP-4-dehydrorhamnose 3,5-epimerase', 20, 'PF00908.hmm'),
    ('Polysacc_synt_2', 'Polysaccharide biosynthesis protein', 20, 'PF02719.hmm'),
    ('RmlD_sub_bind', 'RmlD substrate binding domain', 20, 'PF04321.hmm'),
    ('Polysacc_synt', 'Polysaccharide biosynthesis protein', 23, 'PF01943.hmm'),
    ('Alpha-amylase_C', 'Alpha amylase, C-terminal all-beta domain', 21, 'PF02806.hmm'),
    ('Epimerase_2', 'UDP-N-acetylglucosamine 2-epimerase', 22, 'PF02350.hmm'),
    ('Bac_transf', 'Bacterial sugar transferase', 21, 'PF02397.hmm'),
    ('Wzy_C', 'O-Antigen ligase', 24, 'PF04932.hmm'),
    ('Glyco_transf_9', 'Glycosyltransferase family 9 (heptosyltransferase)', 20, 'PF01075.hmm'),
    ('Glycos_transf_4', 'Glycosyl transferase family 4', 21, 'PF00953.hmm'),
    ('MannoseP_isomer', 'Mannose-6-phosphate isomerase', 23, 'PF01050.hmm'),
    ('Glyco_transf_8', 'Glycosyl transferase family 8', 24, 'PF01501.hmm'),
    ('Capsule_synth', 'Capsule polysaccharide biosynthesis protein', 20, 'PF05159.hmm'),
    ('Glyco_tran_28_C', 'Glycosyltransferase family 28 C-terminal domain', 21, 'PF04101.hmm'),
    ('Poly_export', 'Polysaccharide biosynthesis/export protein', 26, 'PF02563.hmm'),
    ('Glyco_transf_8C', 'Glycosyl transferase family 8 C-terminal', 19, 'PF08437.hmm'),
    ('PIG-L', 'GlcNAc-PI de-N-acetylase', 22, 'PF02585.hmm'),
    ('APE_KS1', 'Aryl polyene ketosynthase 1', 50, 'APE_KS1.hmm'),
    ('APE_KS2', 'Aryl polyene ketosynthase 2', 50, 'APE_KS2.hmm'),
    ('DarB', '(Dialkyl)resorcinol-associated ketosynthase', 50, 'DarB.hmm'),
    ('FabF', 'FabF', 50, 'FabF.hmm'),
    ('ksIII', 'CerJ-like KSIII enzyme', 50, 'ksIII.hmm'),
    ('hr-t2pks-ksa', 'Highly reducing PKS type II ketosynthase alpha', 200, 'hr-t2pks-ksa.hmm'),
    ('ketoacyl-synt', 'Beta-ketoacyl synthase, N-terminal domain', 20, 'PF00109.hmm'),
    ('PUFA_KS', 'Polyunsaturated fatty acid ketosynthase', 50, 'PUFA_KS.hmm'),
    ('PF13471', 'Transglut_core3 domain', 22, 'PF13471.hmm'),
    ('Asn_synthase', 'Asparagine synthase', 21, 'PF00733.hmm'),
    ('PF05402', 'PqqD / RRE protein', 23, 'PF05402.hmm'),
    ('orf2_PTase', 'Orf2-type phenol/phenazine prenyltransferase', 50, 'orf2_PTase.hmm'),
    ('indole_PTase', 'bacterial indole prenyltransferase', 50, 'indole_PTase.hmm'),
    ('ubiA', 'UbiA prenyltransferase of lipoquinone biosynthesis', 50, 'UbiA.hmm'),
    ('pgm1', 'Pgm1-like peptide ligase', 300, 'pgm1.hmm'),
    ('pgm_amidino', 'Pgm-related amidinotransferase', 300, 'pgm_amidino.hmm'),
    ('PF00881', 'TOMM dehydrogenase', 20, 'PF00881.hmm'),
    ('Abhydrolase_6', 'Alpha/beta hydrolase family', 27, 'PF12697.hmm'),
    ('PF00561', 'Abhydrolase', 20, 'PF00561.hmm'),
    ('TIGR04404', 'SkfA: sporulation killing factor', 45, 'TIGR04404.hmm'),
    ('Peptidase_S8', 'Subtilase family', 21, 'PF00082.hmm'),
    ('Peptidase_M16', 'Insulinase (Peptidase family M16)', 22, 'PF00675.hmm'),
    ('Peptidase_S41', 'Peptidase family S41', 21, 'PF03572.hmm'),
    ('Peptidase_M16_C', 'Peptidase M16 inactive domain', 20, 'PF05193.hmm'),
    ('Peptidase_S28', 'Peptidase S28 serine carboxypeptiase', 20, 'Peptidase_S28.hmm'),
    ('Fer4_12', '4Fe-4S single cluster domain', 22, 'PF13353.hmm'),
    ('Abi', 'CAAX protease self-immunity', 21, 'PF02517.hmm'),
    ('Peptidase_M50', 'Peptidase family M50', 27, 'PF02163.hmm'),
    ('Peptidase_S9', 'Prolyl oligopeptidase family', 22, 'PF00326.hmm'),
    ('Peptidase_C39', 'Peptidase C39 family', 21, 'PF03412.hmm'),
    ('ThiF', 'ThiF family', 22, 'PF00899.hmm'),
    ('NasY', 'Long-chain N-acyl amino acid synthase', 20, 'NasY.hmm'),
    ('FMO-like', 'Flavin-dependent monooxygenase', 150, 'PF00743.hmm'),
    ('DUF98', 'p-hydroxybenzoic acid synthase', 21, 'PF01947.hmm'),
    ('PpyS', 'PPY synthase', 300, 'ppys.hmm'),
    ('HMGL-like', 'HMGL-like Pyruvate carboxylase', 30, 'PF00682.hmm'),
    ('RimK', 'RimK-like ATP-grasp domain', 21, 'PF08443.hmm'),
    ('TIGR04424', 'Microcin B17-processing protein McbB', 65, 'TIGR04424.hmm'),
    ('TIGR01716', 'RGG quorum sensing regulator', 37, 'TIGR01716.hmm'),
    ('TP_methylase', 'Tetrapyrrole (Corrin/Porphyrin) methylases', 28, 'TP_methylase.hmm'),
    ('DUF3328', 'UstYa/UstYb-like monooxygenase', 25, 'DUF3328.hmm'),
    ('pop', 'fungal prolyl oligopeptidase', 950, 'pop.hmm'),
    ('ustH', 'fungal ustH-type gamma peptidase', 450, 'ustH.hmm'),
    ('truD', 'TruD pseudouridine synthase', 120, 'truD.hmm'),
    ('micKC', 'Lipolanthine biosynthesis protein', 400, 'micKC.hmm'),
    ('GATase_7', 'Glutamine amidotransferase domain', 25, 'PF13537.hmm'),
    ('TIGR03103', 'GNAT-family acetyltransferase', 569, 'TIGR03103.hmm'),
    ('TIGR03104', 'asparagine synthase family amidotransferase', 570, 'TIGR03104.hmm'),
    ('TIGR03106', 'hydrolase, peptidase M42 family', 398, 'TIGR03106.hmm'),
    ('Peptidase_M42', 'M42 glutamyl aminopeptidase', 20, 'PF05343.hmm'),
    ('CDPS', 'Cyclodipeptide synthase', 27, 'PF16715.hmm'),
    ('2OG-FeII_Oxy', '2OG-Fe(II) oxygenase superfamily', 22, 'PF03171.hmm'),
    ('ATP-grasp', 'ATP-grasp domain', 21, 'PF02222.hmm'),
    ('Chor_lyase', 'Chorismate lyase', 24, 'PF04345.hmm'),
    ('Orn_DAP_Arg_deC', 'Pyridoxal-dependent decarboxylase, C-terminal sheet domain', 21, 'PF00278.hmm'),
    ('Orn_Arg_deC_N', 'Pyridoxal-dependent decarboxylase, pyridoxal binding domain', 24, 'PF02784.hmm'),
    ('GST_N_3', 'Glutathione S-transferase, N-terminal domain', 22, 'PF13417.hmm'),
    ('GST_C', 'Glutathione S-transferase, C-terminal domain', 21, 'PF00043.hmm'),
    ('Acyl-CoA_dh_N', 'Acyl-CoA dehydrogenase, N-terminal domain', 23, 'PF02771.hmm'),
    ('Acyl-CoA_dh_M', 'Acyl-CoA dehydrogenase, middle domain', 33, 'PF02770.hmm'),
    ('Acyl-CoA_dh_1', 'Acyl-CoA dehydrogenase, C-terminal domain', 22, 'PF00441.hmm'),
    ('TIGR02278', 'PaaN-DH: phenylacetic acid degradation protein paaN', 563, 'TIGR02278.hmm'),
    ('TfuA', 'TfuA-like protein', 26, 'PF07812.hmm'),
    ('HopA1', 'HopA1 effector protein family', 23, 'PF17914.hmm'),
    ('APH', 'Phosphotransferase enzyme family', 23, 'PF01636.hmm'),
    ('PPDK_N', 'Pyruvate phosphate dikinase, AMP/ATP binding domain', 20, 'PF01326.hmm'),
    ('TIGR02353', 'Non-alpha poly amino acid condensation domain', 423, 'TIGR02353.hmm'),
    ('Aminotran_3', 'Aminotransferase class-III', 31, 'PF00202.hmm'),
    ('DXP_synthase_N', '1-deoxy-D-xylulose-5-phosphate synthase', 21, 'PF13292.hmm'),
    ('Mac', 'Maltose acetyltransferase', 22, 'PF12464.hmm'),
    ('TIGR03973', 'SCIFF six Cys in fourty-five ranthipeptide precursor', 50, 'TIGR03973.hmm'),
    ('SCIFF', 'Six-cysteine peptide SCIFF', 22, 'PF13165.hmm'),
    ('TIGR04085', 'radical SAM additional 4Fe4S-binding SPASM domain', 27, 'TIGR04085.hmm'),
    ('SPASM', 'Iron-sulfur cluster-binding domain', 21, 'PF13186.hmm'),
    ('TIGR02111', 'coenzyme PQQ biosynthesis protein C', 240, 'TIGR02111.hmm'),
    ('TIGR03966', 'heme/flavin dehydrogenase, mycofactocin system', 471, 'TIGR03966.hmm'),
    ('TIGR04078', 'peptide modification radical SAM enzyme, YydG family', 300, 'TIGR04078.hmm'),
    ('rSAM_GlyRichRpt', 'GRRM system radical SAM/SPASM domain protein', 310, 'TIGR04261.hmm'),
    ('SAM_SPASM_FxsB', 'FxsB family radical SAM/SPASM domain protein', 370, 'TIGR04269.hmm'),
    ('rSAM_XyeB', 'XyeB family radical SAM/SPASM peptide maturase', 450, 'TIGR04496.hmm'),
    ('rSAM_pep_methan', 'TIGR04083 family peptide-modifying radical SAM enzyme', 100, 'TIGR04083.hmm'),
    ('PTHR43273', 'radical SAM/SPASM domain protein, sulfatase-maturase type', 100, 'PTHR43273.hmm'),
    ('methano_modCys', 'TIGR04165 family Cys-rich peptide', 60, 'TIGR04165.hmm'),
    ('yydH', 'Epipeptide metalloprotease yydH', 400, 'yydH.hmm'),
    ('TIGR04223', 'cyclic lactone autoinducer peptide', 19, 'TIGR04223.hmm'),
    ('AgrB', 'cyclic lactone autoinducer endopeptidase', 23, 'PF04647.hmm'),
    ('TIGR04103', 'nif11-class peptide radical SAM maturase 3', 725, 'TIGR04103.hmm'),
    ('Methyltransf_3', 'O-methyltransferase', 22, 'PF01596.hmm'),
    ('DHQ_synthase', '3-dehydroquinate synthase', 24, 'PF01761.hmm'),
    ('Dala_Dala_lig_N', 'D-ala D-ala ligase N-terminus', 34, 'PF01820.hmm'),
    ('Dala_Dala_lig_C', 'D-ala D-ala ligase C-terminus', 24, 'PF07478.hmm'),
    ('ATP-grasp_3', 'ATP-grasp domain', 23, 'PF02655.hmm'),
    ('Carbam_trans_N', 'Carbamoyltransferase N-terminus', 24, 'PF02543.hmm'),
    ('PhyH', 'Phytanoyl-CoA dioxygenase (PhyH)', 20, 'PF05721.hmm'),
    ('ADSI-DH', '3-amino-2,3-dideoxy-scyllo-inositol 1-dehydrogenase', 460, 'adsi_dh.hmm'),
    ('ADH_N', 'Alcohol dehydrogenase GroES-like domain', 21, 'PF08240.hmm'),
    ('ADH_zinc_N', 'Zinc-binding dehydrogenase', 32, 'PF00107.hmm'),
    ('RCDPS', 'arginine-containing cyclic dipeptide synthase', 500, 'rcdps.hmm'),
    ('DIT1_PvcA', 'Pyoverdine/dityrosine biosynthesis protein', 27, 'PF05141.hmm'),
    ('TauD', 'Taurine catabolism dioxygenase TauD, TfdA family', 27, 'PF02668.hmm'),
    ('Transferase', 'Transferase family', 20, 'PF02458.hmm'),
    ('Fer2_4', '2Fe-2S iron-sulfur cluster binding domain', 22, 'PF13510.hmm'),
    ('Fer2_BFD', 'BFD-like [2Fe-2S] binding domain', 23, 'PF04324.hmm'),
    ('Pyr_redox_2', 'Pyridine nucleotide-disulphide oxidoreductase', 26, 'PF07992.hmm'),
    ('DAO', 'FAD dependent oxidoreductase', 27, 'PF01266.hmm'),
    ('Lanthipeptide_LanB_RRE', 'RRE-containing LanB protein in a class I lanthipeptide cluster', 25, 'RREFam001.hmm'),
    ('Stand_Alone_Lasso_RRE', 'Stand-alone RRE protein in a lasso peptide cluster', 25, 'RREFam002.hmm'),
    ('Lasso_Fused_RRE', 'RRE-containing fused peptidase in a lasso peptide cluster', 25, 'RREFam003.hmm'),
    ('Thiopeptide_F_RRE', 'RRE-containing F protein in a thiopeptide cluster', 25, 'RREFam004.hmm'),
    ('PqqD_RRE', 'RRE-containing protein in a pyrroloquinoline cluster', 25, 'RREFam006.hmm'),
    ('Cyanobactin_D_RRE', 'RRE-containing D protein in a cyanobactin cluster', 25, 'RREFam007.hmm'),
    ('Bottromycin_Methyltransferase_RRE', 'RRE-containing methyltransferase protein in a bottromycin cluster', 25, 'RREFam008.hmm'),
    ('Proteusin_Methyltransferase_RRE', 'RRE-containing methyltransferase in a proteusin cluster', 25, 'RREFam009.hmm'),
    ('Proteusin_Epimerase_RRE', 'RRE-containing epimerase in a proteusin cluster', 25, 'RREFam010.hmm'),
    ('Heterocycloanthracin_C_RRE', 'RRE-containing C protein in a heterocycloanthracin LAP cluster', 25, 'RREFam011.hmm'),
    ('Plantazolicin_C_RRE', 'RRE-containing discrete C protein in a plantazolicin LAP cluster', 25, 'RREFam012.hmm'),
    ('Goadsporin_CD_RRE', 'RRE-containing CD fusion protein in a goadsporin cluster', 25, 'RREFam013.hmm'),
    ('Cytolysin_C_RRE', 'RRE-containing discrete C protein in a SLS/cytolysin LAP cluster', 25, 'RREFam014.hmm'),
    ('NHLP_CD_RRE', 'RRE-containing CD fusion protein in a NHLP/Nif11 LAP cluster', 25, 'RREFam015.hmm'),
    ('Mycofactocin_RRE', 'Stand-alone RRE protein in a mycofactocin cluster', 25, 'RREFam016.hmm'),
    ('Pantocin_Microcin_RRE', 'RRE-containing E1-like protein in a pantocin/microcin cluster', 25, 'RREFam017.hmm'),
    ('Subtilosin_rSAM_RRE', 'RRE-containing rSAM protein in a subtilosin sactipeptide cluster', 25, 'RREFam018.hmm'),
    ('Thurincin_rSAM_RRE', 'RRE-containing rSAM protein in a thurincin sactipeptide cluster', 25, 'RREFam019.hmm'),
    ('Thuricin_rSAM_RRE', 'RRE-containing rSAM protein in a thuricin sactipeptide cluster', 25, 'RREFam020.hmm'),
    ('Other_Sactipeptide_rSAM_RRE', 'RRE-containing rSAM protein in a sactipeptide cluster', 25, 'RREFam021.hmm'),
    ('Streptide_RRE', 'RRE-containing rSAM enzyme in a streptide-like cluster', 25, 'RREFam022.hmm'),
    ('Quinohemoprotein_rSAM_RRE', 'RRE-containing rSAM protein in a quinohemoprotein cluster', 25, 'RREFam023.hmm'),
    ('Ranthipeptide_rSAM_RRE', 'RRE-containing rSAM protein in a ranthipeptide/SCIFF cluster', 25, 'RREFam024.hmm'),
    ('Trifolitoxin_RRE', 'RRE-containing protein in a trifolitoxin cluster', 25, 'RREFam025.hmm'),
    ('Thiaglutamate_G_RRE', 'RRE-peptidase fusion protein in a 3-thiaglutamate cluster', 25, 'RREFam026.hmm'),
    ('Thiaglutamate_I_RRE', 'RRE-DUF fusion protein in a 3-thiaglutamate cluster', 25, 'RREFam027.hmm'),
    ('Thiaglutamate_B_RRE', 'RRE-LanB fusion protein in a 3-thiaglutamate cluster', 25, 'RREFam028.hmm'),
    ('PlpY_RRE', 'Discrete RRE protein in an alpha-keto beta-amino acid containing peptide cluster', 25, 'RREFam029.hmm'),
    ('Cy_tandem', 'Duplicated cyclization domain found in siderophore clusters', 350, 'Cy_tandem.hmm'),
    ('EntA', '2,3-dihydro-2,3-dihydroxybenzoic acid dehydrogenase', 205, 'EntA.hmm'),
    ('EntC', 'Isochorismate synthase', 300, 'EntC.hmm'),
    ('FbnL', 'DMAQ biosynthesis protein found in fabrubactin cluster', 40, 'FbnL.hmm'),
    ('FbnM', 'DMAQ biosynthesis protein found in fabrubactin cluster', 180, 'FbnM.hmm'),
    ('GrbD', 'Graminine biosynthesis protein found in gramibactin cluster', 400, 'GrbD.hmm'),
    ('GrbE', 'Graminine biosynthesis protein found in gramibactin cluster', 250, 'GrbE.hmm'),
    ('IBH_Asp', 'Interface-associated aspartate beta-hydroxylase', 440, 'IBH_Asp.hmm'),
    ('IBH_His', 'Interface-associated histidine beta-hydroxylase', 470, 'IBH_His.hmm'),
    ('SBH_Asp', 'Syringomycin-type aspartate beta-hydroxylase not in siderophores', 500, 'SBH_Asp.hmm'),
    ('TBH_Asp', 'T_E-associated aspartate beta-hydroxylase', 420, 'TBH_Asp.hmm'),
    ('CyanoBH_Asp1', 'Cyanobacterial siderophore Asp beta-hydroxylase (NRPS domain)', 400, 'CyanoBH_Asp1.hmm'),
    ('CyanoBH_Asp2', 'Cyanobacterial siderophore Asp beta-hydroxylase (standalone)', 450, 'CyanoBH_Asp2.hmm'),
    ('IPL', 'Isopyruvate lyase', 110, 'IPL.hmm'),
    ('Lys_monoox', 'Lysine N6-monooxygenase (Mycobactin-type)', 300, 'Lys_monoox.hmm'),
    ('Orn_monoox', 'Ornithine N5-monooxygenase (Pyoverdine-type)', 415, 'Orn_monoox.hmm'),
    ('KtzT', 'Heme-dependent piperazate synthase (10.1038/nchembio.2411) not in siderophores', 200, 'KtzT.hmm'),
    ('MetRS-like', 'Methionyl-tRNA synthase-like ligase (10.1038/s41467-021-27523-x) not in siderophores', 360, 'MetRS-like.hmm'),
    ('PvdO', 'Oxidoreductase in pyoverdine chromophore maturation', 250, 'PvdO.hmm'),
    ('PvdP', 'Tyrosinase enzyme in pyoverdine chromophore maturation', 600, 'PvdP.hmm'),
    ('SalSyn', 'Salicylate synthase', 350, 'SalSyn.hmm'),
    ('VbsL', 'PLP-dependent epimerase in vicibactin biosynthesis (10.1021/ja9056008)', 570, 'VbsL.hmm'),
    ('VibH_like', 'VibH-like standalone condensation domain', 400, 'VibH_like.hmm'),
    ('MbnC', 'Methanobactin maturation protein MbnC', 150, 'MbnC.hmm'),
    ('CntL', 'NAS synthase found in opine-like zincophore clusters', 100, 'CntL.hmm'),
    ('CntM', 'Staphylopine dehydrogenase found in opine-like zincophore clusters', 100, 'CntM.hmm'),
    ('AesA', 'N-((2S)-2-amino-2-carboxyethyl)-L-glutamate dehydrogenase involved in APCA biosynthesis', 300, 'AesA.hmm'),
    ('AesB', 'Pyridoxal-dependent decarboxylase involved in APCA biosynthesis', 400, 'AesB.hmm'),
    ('AesC', 'N-(2-amino-2-carboxyethyl)-L-glutamate synthase involved in APCA biosynthesis', 420, 'AesC.hmm'),
    ('4HBT_2', 'Hotdog fold Thioesterase-like superfamily', 25, 'PF13279.hmm'),
    ('CDPS_fung', 'Fungal cyclodipeptide synthase', 100, 'CDPS_fung.hmm'),
    ('GTP_cyclohydroI', 'GTP cyclohydrolase I', 27, 'PF01227.hmm'),
    ('PTPS', '6-carboxy-5,6,7,8-tetrahydropterin synthase', 22, 'PF01242.hmm'),
    ('QueE', '7-carboxy-7-deazaguanine synthase', 22, 'QueE.hmm'),
    ('QueC', '7-cyano-7-deazaguanine synthase', 22, 'PF06508.hmm'),
    ('dpdA1', 'family of paralogs of tRNA guanosine transglycosylase genes', 20, 'dpdA1.hmm'),
    ('alpha_am_amid', 'L-aminoadipate-semialdehyde dehydrogenase', 1030, 'TIGR03443.hmm'),
    ('bTGT', 'homodimeric bacterial tRNA-guanine transglycosylases', 22, 'bTGT.hmm'),
    ('delta9_cd03505', 'delta9 desaturase', 22, 'delta9_cd03505.hmm'),
    ('delta12_cd03507', 'delta12 desaturase', 22, 'delta12_cd03507.hmm'),
    ('FAAL_cd05931', 'FAAL', 500, 'FAAL_cd05931.hmm'),
    ('PP-binding_2', 'Acyl-carrier', 10, 'PF14573.hmm'),
    ('Thioesterase', 'Thioesterase domain', 26, 'PF00975.hmm'),
    ('LPG_synthase_C', 'Phosphatidylglycerol lysyltransferase, C-terminal', 27, 'PF09924.hmm'),
    ('azdH', 'azdH', 50, 'azdh.hmm'),
    ('azdO', 'azdO/vlmO', 50, 'azdO.hmm'),
    ('Ketoacyl-synt_2', 'Ketoacyl-synt_2', 25, 'PF13723.hmm'),
    ('PT_phytoene_like', 'Prenyltransferase; Phytoene synthase like', 20, 'PT_phytoene_like.hmm'),
    ('PT_FPPS_like', 'Prenyltransferase; Farnesyl diphosphate synthase like', 56, 'PT_FPPS_like.hmm'),
    ('PT_HPP', 'Prenyltransferase; involved in isoprenoid quinone biosynthesis', 139, 'PT_HPP.hmm'),
    ('T1TS', 'Type 1 terpene synthase', 15, 'T1TS.hmm'),
    ('T1TS_KS', 'Type 1 terpene synthase; Ent-kaurene synthase type', 200, 'T1TS_KS.hmm'),
    ('T2TS', 'Type 2 terpene synthase', 45, 'T2TS.hmm'),
    ('TS_UbiA', 'UbiA type terpene synthase', 117, 'TS_UbiA.hmm'),
    ('HAD_2', 'Haloacid dehalogenase-like hydrolase', 15, 'PF13419.hmm'),
    ('TS_Pyr4', 'Pyr4 type terpene synthase', 50, 'TS_Pyr4.hmm'),
    ('Lycopene_cycl_fung', 'Lycopene cyclase domain; fungal', 40, 'TIGR03462.hmm'),
    ('IPPT', 'IPP transferase', 24, 'PF01715.hmm'),
    ('Lysine_decarbox', 'Possible lysine decarboxylase', 27, 'PF03641.hmm'),
    ('Amidohydro_1', 'Amidohydrolase family', 25, 'PF01979.hmm'),
    ('GFO_IDH_MocA', 'Oxidoreductase family, NAD-binding Rossmann fold', 24, 'PF01408.hmm'),
    ('GFO_IDH_MocA_C3', 'GFO/IDH/MocA C-terminal domain', 27, 'PF22725.hmm'),
    ('SIS_2', 'SIS domain', 27, 'PF13580.hmm'),
    ('her4', 'NAD-dependent S-adenosylhomocysteine (SAH) hydrolase', 400, 'her4.hmm'),
    ('her5', 'NAD-binding oxidoreductase', 200, 'her5.hmm'),
    ('her6', 'B12-dependent radical S-adenosyl-l-methionine enzyme', 400, 'her6.hmm'),
    ('her7', 'NAD-binding oxidoreductase', 550, 'her7.hmm'),
    ('atropopeptide_precursor', 'Pattern-based detection of atropopeptide precursor peptides', NULL, 'atropopeptide_precursor.py'),
    ('cyanobactin_precursor', 'Pattern-based detection of cyanobactin precursor peptides', NULL, 'cyanobactin_precursor.py'),
    ('darobactin_precursor', 'Pattern-based detection of darobactin precursor peptides', NULL, 'darobactin_precursor.py'),
    ('darobactin_rSAM', 'Detection of darobactin-related rSAM based on precursor presence', NULL, 'darobactin_rsam.py'),
    ('triceptide_precursor', 'Pattern-based detection of triceptide precursor peptides', NULL, 'triceptide_precursor.py'),
    ('triceptide_rSAM', 'Detection of triceptide-related rSAM based on precursor presence', NULL, 'triceptide_rsam.py');