annotate lexmapr/definitions.py @ 0:f5c39d0447be

"planemo upload"
author kkonganti
date Wed, 31 Aug 2022 14:32:07 -0400
parents
children
rev   line source
kkonganti@0 1 """Static definitions"""
kkonganti@0 2
kkonganti@0 3 import os
kkonganti@0 4
kkonganti@0 5
kkonganti@0 6 # root path
kkonganti@0 7 ROOT = os.path.dirname(__file__)
kkonganti@0 8
kkonganti@0 9 # URL to list of OLS ontologies where download link is given
kkonganti@0 10 embl_ontologies = 'https://www.ebi.ac.uk/ols/ontologies'
kkonganti@0 11
kkonganti@0 12 # beginning of URL to ontology PURL
kkonganti@0 13 purl_link = 'http://purl.obolibrary.org/obo/'
kkonganti@0 14
kkonganti@0 15 # directory for downloaded OWL files
kkonganti@0 16 owl_dir = 'lexmapr/owl_files'
kkonganti@0 17
kkonganti@0 18 # path to database with synonyms from predefined resources and from OWL files
kkonganti@0 19 synonym_db = 'lexmapr/owl_files/label_synonyms.db'
kkonganti@0 20
kkonganti@0 21 # path to database with all ontologies of interest
kkonganti@0 22 ontol_db = 'lexmapr/owl_files/ontol_table.db'
kkonganti@0 23
kkonganti@0 24
kkonganti@0 25 # ontologies of interest
kkonganti@0 26 ontol_interest = [#'BFO',
kkonganti@0 27 #'CHEBI',
kkonganti@0 28 #'ENVO',
kkonganti@0 29 'FOODON',
kkonganti@0 30 #'GENEPIO',
kkonganti@0 31 'NCBITAXON', # NCBITaxon is not valid as of April 2022
kkonganti@0 32 #'OGMS',
kkonganti@0 33 #'PATO',
kkonganti@0 34 #'PCO',
kkonganti@0 35 #'UBERON',
kkonganti@0 36 ]
kkonganti@0 37
kkonganti@0 38 # ontology accessions that do not have labels or are placeholders as of April 2022
kkonganti@0 39 # will skip in database building
kkonganti@0 40 missing_ontol_labels = ['GENEPIO_0001367','GENEPIO_0001368','GENEPIO_0001369','GENEPIO_0001370',
kkonganti@0 41 'GENEPIO_0001372','GENEPIO_0001373','_MIAA_0000021',
kkonganti@0 42 ]
kkonganti@0 43
kkonganti@0 44 # terms indicating that the metadata was not given/collected; will output empty results
kkonganti@0 45 not_provided = ['not applicable','unknown','n a','not provided','not available','miscellaneous',
kkonganti@0 46 'not collected','missing','unidentified','unknown','none','unamed','other',
kkonganti@0 47 'undetermined','not known','no history given','no source specified','null',
kkonganti@0 48 'unspecified','not reported','not available not collected','not isolated',
kkonganti@0 49 'not available','not provided','xxx','mising','misng','other','unidentified',
kkonganti@0 50 'not determined other','reported later','intact unknown','not determined',
kkonganti@0 51 'not ascertained','unk','nd','nd others','nd other','etc','na','',' ',
kkonganti@0 52 'not supplied','not specified',
kkonganti@0 53 ]
kkonganti@0 54
kkonganti@0 55 # below are bin definitions
kkonganti@0 56 # TODO: food consumer group:FOODON_03510136 changed, formatted as '* as food consumer'
kkonganti@0 57 # can collect as 'for *' in text?
kkonganti@0 58 #fo_consumer = []
kkonganti@0 59
kkonganti@0 60 fo_product = ['algal food product:FOODON_00001184',
kkonganti@0 61 'amphibian:FOODON_03411624',
kkonganti@0 62 'amphibian or reptile food product:FOODON_00002200',
kkonganti@0 63 'animal based refined or partially-refined food product:FOODON_00001595',
kkonganti@0 64 'avian egg food product:FOODON_00001105',
kkonganti@0 65 'avian food product:FOODON_001251',
kkonganti@0 66 'bakery food product:FOODON_00001626',
kkonganti@0 67 'cell-based technology food product:FOODON_00003376',
kkonganti@0 68 'dairy food product:FOODON_00001256',
kkonganti@0 69 'dietary supplement:FOODON_03401298',
kkonganti@0 70 'fish egg food product:FOODON_00001250',
kkonganti@0 71 'fish food product:FOODON_00001248',
kkonganti@0 72 'food product analog:FOODON_00001871',
kkonganti@0 73 'food product component:FOODON_00001714',
kkonganti@0 74 'fungus food product:FOODON_00001143',
kkonganti@0 75 'game animal food product:FOODON_00002477',
kkonganti@0 76 'insect food product:FOODON_00001177',
kkonganti@0 77 'meat food product:FOODON_00002477',
kkonganti@0 78 'microbial food product:FOODON_00001145',
kkonganti@0 79 'plant food product:FOODON_00001015',
kkonganti@0 80 'poultry food product:FOODON_00001283',
kkonganti@0 81 'prepared food product:FOODON_00001180',
kkonganti@0 82 'processed food product:FOODON_03311737',
kkonganti@0 83 'reptile egg food product:FOODON_00002199',
kkonganti@0 84 'seafood product:FOODON_00001046',
kkonganti@0 85 'shellfish food product:FOODON_00001293',
kkonganti@0 86 'soup food product:FOODON_00002257',
kkonganti@0 87 'sustainable agriculture food product:FOODON_00003375',
kkonganti@0 88 'vegetarian food product:FOODON_00003194',
kkonganti@0 89 'vertebrate animal food product:FOODON_00001092',
kkonganti@0 90 ]
kkonganti@0 91
kkonganti@0 92 fo_quality = ['food (acidified):FOODON_03301625',
kkonganti@0 93 'food (adulterated):FOODON_00003367',
kkonganti@0 94 'food (baked):FOODON_00002456',
kkonganti@0 95 'food (batter-coated):FOODON_00002662',
kkonganti@0 96 'food (blanched):FOODON_00002767',
kkonganti@0 97 'food (blend):FOODON_00003889',
kkonganti@0 98 'food (boiled):FOODON_00002688',
kkonganti@0 99 'food (breaded):FOODON_00002661',
kkonganti@0 100 'food (broiled or grilled):FOODON_00002647',
kkonganti@0 101 'food (canned):FOODON_00002418',
kkonganti@0 102 'food (chilled):FOODON_00002642',
kkonganti@0 103 'food (chopped):FOODON_00002777',
kkonganti@0 104 'food (cleaned):FOODON_00002708',
kkonganti@0 105 'food (colored):FOODON_00002650',
kkonganti@0 106 'food (comminuted):FOODON_00002754',
kkonganti@0 107 'food (cooked):FOODON_00001181',
kkonganti@0 108 'food (deep-fried):FOODON_03307052',
kkonganti@0 109 'food (dehydrated):FOODON_00002643',
kkonganti@0 110 'food (dried):FOODON_03307539',
kkonganti@0 111 'food (fat or oil coated):FOODON_03460233',
kkonganti@0 112 'food (fermented):FOODON_00001258',
kkonganti@0 113 'food (filled):FOODON_00002644',
kkonganti@0 114 'food (flavored):FOODON_00002646',
kkonganti@0 115 'food (freeze-dried):FOODON_03301752',
kkonganti@0 116 'food (fresh):FOODON_00002457',
kkonganti@0 117 'food (fried):FOODON_00002660',
kkonganti@0 118 'food (frozen):FOODON_03302148',
kkonganti@0 119 'food (genetically-modified):FOODON_03530251',
kkonganti@0 120 'food (ground):FOODON_00002713',
kkonganti@0 121 'food (harvested):FOODON_00003398',
kkonganti@0 122 'food (heat treated):FOODON_03316043',
kkonganti@0 123 'food (hulled):FOODON_00002720',
kkonganti@0 124 'food (hydrolized):FOODON_00002653',
kkonganti@0 125 'food (irradiated):FOODON_03305364',
kkonganti@0 126 'food (juiced):FOODON_00003499',
kkonganti@0 127 'food (liquid):FOODON_03430130',
kkonganti@0 128 'food (milled):FOODON_00002649',
kkonganti@0 129 'food (not genetically-modified):FOODON_00003379',
kkonganti@0 130 'food (organically grown):FOODON_03306690',
kkonganti@0 131 'food (packaged):FOODON_00002739',
kkonganti@0 132 'food (packed in high pressurised containers):FOODON_03317139',
kkonganti@0 133 'food (pan-fried):FOODON_00002463',
kkonganti@0 134 'food (paste):FOODON_00003887',
kkonganti@0 135 'food (pasteurized):FOODON_00002654',
kkonganti@0 136 'food (peeled):FOODON_00002655',
kkonganti@0 137 'food (pickled):FOODON_00001079',
kkonganti@0 138 'food (powdered):FOODON_00002976',
kkonganti@0 139 'food (precooked):FOODON_00002971',
kkonganti@0 140 'food (precooked, frozen):FOODON_03305323',
kkonganti@0 141 'food (preserved):FOODON_00002158',
kkonganti@0 142 'food (puffed):FOODON_00002656',
kkonganti@0 143 'food (raw):FOODON_03311126',
kkonganti@0 144 'food (rehydrated):FOODON_00002755',
kkonganti@0 145 'food (roasted):FOODON_00002744',
kkonganti@0 146 'food (salted):FOODON_03460173',
kkonganti@0 147 'food (seasoned):FOODON_00002733',
kkonganti@0 148 'fruit (seedless):FOODON_00003461',
kkonganti@0 149 'food (semiliquid):FOODON_03430103',
kkonganti@0 150 'food (semisolid):FOODON_03430144',
kkonganti@0 151 'food (sliced):FOODON_00002455',
kkonganti@0 152 'food (smoked or smoke-flavored):FOODON_03460172',
kkonganti@0 153 'food (solid):FOODON_03430151',
kkonganti@0 154 'food (spoiled):FOODON_00003366',
kkonganti@0 155 'food (starch or flour thickened):FOODON_03315268',
kkonganti@0 156 'food (steamed):FOODON_00002657',
kkonganti@0 157 'food (sugar-free):FOODON_03315838',
kkonganti@0 158 'food (textured):FOODON_00002658',
kkonganti@0 159 'food (toasted):FOODON_00002659',
kkonganti@0 160 'food (unprocessed):FOODON_03316056',
kkonganti@0 161 'food (unstandardized):FOODON_03315636',
kkonganti@0 162 ]
kkonganti@0 163
kkonganti@0 164 fo_organism = ['algae:FOODON_03411301',
kkonganti@0 165 'animal:FOODON_00003004',
kkonganti@0 166 'fungus:FOODON_03411261',
kkonganti@0 167 'lichen:FOODON_03412345',
kkonganti@0 168 'whole plant:PO_0000003',
kkonganti@0 169 ]
kkonganti@0 170
kkonganti@0 171 ncbi_taxon = ['Actinopterygii:NCBITaxon_7898', #mix of taxon types
kkonganti@0 172 'Ecdysozoa:NCBITaxon_1206794',
kkonganti@0 173 'Echinodermata:NCBITaxon_7586',
kkonganti@0 174 'Fungi:NCBITaxon_4751',
kkonganti@0 175 'Spiralia:NCBITaxon_2697495',
kkonganti@0 176 'Viridiplantae:NCBITaxon_33090',
kkonganti@0 177 'Amphibia:NCBITaxon_8292',
kkonganti@0 178 #'Sauropsida:NCBITaxon_8457',
kkonganti@0 179 'Aves:NCBITaxon_8782',
kkonganti@0 180 'Crocodylia:NCBITaxon_1294634',
kkonganti@0 181 'Testudinata:NCBITaxon_2841271',
kkonganti@0 182 'Lepidosauria:NCBITaxon_8504',
kkonganti@0 183 #'Mammalia:NCBITaxon_40674',
kkonganti@0 184 'Artiodactyla:NCBITaxon_91561',
kkonganti@0 185 'Carnivora:NCBITaxon_33554',
kkonganti@0 186 'Chiroptera:NCBITaxon_9397',
kkonganti@0 187 'Chrysochloridae:NCBITaxon_9389',
kkonganti@0 188 'Eulipotyphla:NCBITaxon_9362',
kkonganti@0 189 'Hyracoidea:NCBITaxon_9810',
kkonganti@0 190 'Macroscelidea:NCBITaxon_28734',
kkonganti@0 191 'Metatheria:NCBITaxon_9263',
kkonganti@0 192 'Ornithorhynchidae:NCBITaxon_9256',
kkonganti@0 193 'Perissodactyla:NCBITaxon_9787',
kkonganti@0 194 'Pholidota:NCBITaxon_9971',
kkonganti@0 195 'Primates:NCBITaxon_9443',
kkonganti@0 196 'Proboscidea:NCBITaxon_9779',
kkonganti@0 197 'Rodentia:NCBITaxon_9989',
kkonganti@0 198 'Sirenia:NCBITaxon_9774',
kkonganti@0 199 'Tachyglossidae:NCBITaxon_9259',
kkonganti@0 200 'Tenrecidae:NCBITaxon_9369',
kkonganti@0 201 'Tubulidentata:NCBITaxon_9815',
kkonganti@0 202 'Xenarthra:NCBITaxon_9348',
kkonganti@0 203 ]
kkonganti@0 204
kkonganti@0 205 arg_bins = {#'fo_consumer':fo_consumer,
kkonganti@0 206 'fo_product':fo_product,
kkonganti@0 207 'fo_quality':fo_quality,
kkonganti@0 208 'fo_organism':fo_organism,
kkonganti@0 209 'ncbi_taxon':ncbi_taxon,
kkonganti@0 210 }