cstrittmatter@0: """Static definitions"""
cstrittmatter@0: 
cstrittmatter@0: import os
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: # root path
cstrittmatter@0: ROOT = os.path.dirname(__file__)
cstrittmatter@0: 
cstrittmatter@0: # URL to list of OLS ontologies where download link is given
cstrittmatter@0: embl_ontologies = 'https://www.ebi.ac.uk/ols/ontologies'
cstrittmatter@0: 
cstrittmatter@0: # beginning of URL to ontology PURL
cstrittmatter@0: purl_link = 'http://purl.obolibrary.org/obo/'
cstrittmatter@0: 
cstrittmatter@0: # directory for downloaded OWL files
cstrittmatter@0: owl_dir = 'lexmapr/owl_files'
cstrittmatter@0: 
cstrittmatter@0: # path to database with synonyms from predefined resources and from OWL files
cstrittmatter@0: synonym_db = 'lexmapr/owl_files/label_synonyms.db'
cstrittmatter@0: 
cstrittmatter@0: # path to database with all ontologies of interest
cstrittmatter@0: ontol_db = 'lexmapr/owl_files/ontol_table.db'
cstrittmatter@0: 
cstrittmatter@0: 
cstrittmatter@0: # ontologies of interest
cstrittmatter@0: ontol_interest = [#'BFO',
cstrittmatter@0:                   #'CHEBI',
cstrittmatter@0:                   #'ENVO',
cstrittmatter@0:                   'FOODON',
cstrittmatter@0:                   #'GENEPIO',
cstrittmatter@0:                   'NCBITAXON', # NCBITaxon is not valid as of April 2022
cstrittmatter@0:                   #'OGMS',
cstrittmatter@0:                   #'PATO',
cstrittmatter@0:                   #'PCO',
cstrittmatter@0:                   #'UBERON',
cstrittmatter@0:                  ]
cstrittmatter@0: 
cstrittmatter@0: # ontology accessions that do not have labels or are placeholders as of April 2022
cstrittmatter@0: # will skip in database building
cstrittmatter@0: missing_ontol_labels = ['GENEPIO_0001367','GENEPIO_0001368','GENEPIO_0001369','GENEPIO_0001370',
cstrittmatter@0:                         'GENEPIO_0001372','GENEPIO_0001373','_MIAA_0000021',
cstrittmatter@0:                        ]
cstrittmatter@0: 
cstrittmatter@0: # terms indicating that the metadata was not given/collected; will output empty results
cstrittmatter@0: not_provided = ['not applicable','unknown','n a','not provided','not available','miscellaneous',
cstrittmatter@0:                 'not collected','missing','unidentified','unknown','none','unamed','other',
cstrittmatter@0:                 'undetermined','not known','no history given','no source specified','null',
cstrittmatter@0:                 'unspecified','not reported','not available not collected','not isolated',
cstrittmatter@0:                 'not available','not provided','xxx','mising','misng','other','unidentified',
cstrittmatter@0:                 'not determined other','reported later','intact unknown','not determined',
cstrittmatter@0:                 'not ascertained','unk','nd','nd others','nd other','etc','na','',' ',
cstrittmatter@0:                 'not supplied','not specified',
cstrittmatter@0:                ]
cstrittmatter@0: 
cstrittmatter@0: # below are bin definitions
cstrittmatter@0: # TODO: food consumer group:FOODON_03510136 changed, formatted as '* as food consumer'
cstrittmatter@0: #       can collect as 'for *' in text?
cstrittmatter@0: #fo_consumer = []
cstrittmatter@0: 
cstrittmatter@0: fo_product = ['algal food product:FOODON_00001184',
cstrittmatter@0:               'amphibian:FOODON_03411624',
cstrittmatter@0:               'amphibian or reptile food product:FOODON_00002200',
cstrittmatter@0:               'animal based refined or partially-refined food product:FOODON_00001595',
cstrittmatter@0:               'avian egg food product:FOODON_00001105',
cstrittmatter@0:               'avian food product:FOODON_001251',
cstrittmatter@0:               'bakery food product:FOODON_00001626',
cstrittmatter@0:               'cell-based technology food product:FOODON_00003376',
cstrittmatter@0:               'dairy food product:FOODON_00001256',
cstrittmatter@0:               'dietary supplement:FOODON_03401298',
cstrittmatter@0:               'fish egg food product:FOODON_00001250',
cstrittmatter@0:               'fish food product:FOODON_00001248',
cstrittmatter@0:               'food product analog:FOODON_00001871',
cstrittmatter@0:               'food product component:FOODON_00001714',
cstrittmatter@0:               'fungus food product:FOODON_00001143',
cstrittmatter@0:               'game animal food product:FOODON_00002477',
cstrittmatter@0:               'insect food product:FOODON_00001177',
cstrittmatter@0:               'meat food product:FOODON_00002477',
cstrittmatter@0:               'microbial food product:FOODON_00001145',
cstrittmatter@0:               'plant food product:FOODON_00001015',
cstrittmatter@0:               'poultry food product:FOODON_00001283',
cstrittmatter@0:               'prepared food product:FOODON_00001180',
cstrittmatter@0:               'processed food product:FOODON_03311737',
cstrittmatter@0:               'reptile egg food product:FOODON_00002199',
cstrittmatter@0:               'seafood product:FOODON_00001046',
cstrittmatter@0:               'shellfish food product:FOODON_00001293',
cstrittmatter@0:               'soup food product:FOODON_00002257',
cstrittmatter@0:               'sustainable agriculture food product:FOODON_00003375',
cstrittmatter@0:               'vegetarian food product:FOODON_00003194',
cstrittmatter@0:               'vertebrate animal food product:FOODON_00001092',
cstrittmatter@0:              ]
cstrittmatter@0: 
cstrittmatter@0: fo_quality = ['food (acidified):FOODON_03301625',
cstrittmatter@0:               'food (adulterated):FOODON_00003367',
cstrittmatter@0:               'food (baked):FOODON_00002456',
cstrittmatter@0:               'food (batter-coated):FOODON_00002662',
cstrittmatter@0:               'food (blanched):FOODON_00002767',
cstrittmatter@0:               'food (blend):FOODON_00003889',
cstrittmatter@0:               'food (boiled):FOODON_00002688',
cstrittmatter@0:               'food (breaded):FOODON_00002661',
cstrittmatter@0:               'food (broiled or grilled):FOODON_00002647',
cstrittmatter@0:               'food (canned):FOODON_00002418',
cstrittmatter@0:               'food (chilled):FOODON_00002642',
cstrittmatter@0:               'food (chopped):FOODON_00002777',
cstrittmatter@0:               'food (cleaned):FOODON_00002708',
cstrittmatter@0:               'food (colored):FOODON_00002650',
cstrittmatter@0:               'food (comminuted):FOODON_00002754',
cstrittmatter@0:               'food (cooked):FOODON_00001181',
cstrittmatter@0:               'food (deep-fried):FOODON_03307052',
cstrittmatter@0:               'food (dehydrated):FOODON_00002643',
cstrittmatter@0:               'food (dried):FOODON_03307539',
cstrittmatter@0:               'food (fat or oil coated):FOODON_03460233',
cstrittmatter@0:               'food (fermented):FOODON_00001258',
cstrittmatter@0:               'food (filled):FOODON_00002644',
cstrittmatter@0:               'food (flavored):FOODON_00002646',
cstrittmatter@0:               'food (freeze-dried):FOODON_03301752',
cstrittmatter@0:               'food (fresh):FOODON_00002457',
cstrittmatter@0:               'food (fried):FOODON_00002660',
cstrittmatter@0:               'food (frozen):FOODON_03302148',
cstrittmatter@0:               'food (genetically-modified):FOODON_03530251',
cstrittmatter@0:               'food (ground):FOODON_00002713',
cstrittmatter@0:               'food (harvested):FOODON_00003398',
cstrittmatter@0:               'food (heat treated):FOODON_03316043',
cstrittmatter@0:               'food (hulled):FOODON_00002720',
cstrittmatter@0:               'food (hydrolized):FOODON_00002653',
cstrittmatter@0:               'food (irradiated):FOODON_03305364',
cstrittmatter@0:               'food (juiced):FOODON_00003499',
cstrittmatter@0:               'food (liquid):FOODON_03430130',
cstrittmatter@0:               'food (milled):FOODON_00002649',
cstrittmatter@0:               'food (not genetically-modified):FOODON_00003379',
cstrittmatter@0:               'food (organically grown):FOODON_03306690',
cstrittmatter@0:               'food (packaged):FOODON_00002739',
cstrittmatter@0:               'food (packed in high pressurised containers):FOODON_03317139',
cstrittmatter@0:               'food (pan-fried):FOODON_00002463',
cstrittmatter@0:               'food (paste):FOODON_00003887',
cstrittmatter@0:               'food (pasteurized):FOODON_00002654',
cstrittmatter@0:               'food (peeled):FOODON_00002655',
cstrittmatter@0:               'food (pickled):FOODON_00001079',
cstrittmatter@0:               'food (powdered):FOODON_00002976',
cstrittmatter@0:               'food (precooked):FOODON_00002971',
cstrittmatter@0:               'food (precooked, frozen):FOODON_03305323',
cstrittmatter@0:               'food (preserved):FOODON_00002158',
cstrittmatter@0:               'food (puffed):FOODON_00002656',
cstrittmatter@0:               'food (raw):FOODON_03311126',
cstrittmatter@0:               'food (rehydrated):FOODON_00002755',
cstrittmatter@0:               'food (roasted):FOODON_00002744',
cstrittmatter@0:               'food (salted):FOODON_03460173',
cstrittmatter@0:               'food (seasoned):FOODON_00002733',
cstrittmatter@0:               'fruit (seedless):FOODON_00003461',
cstrittmatter@0:               'food (semiliquid):FOODON_03430103',
cstrittmatter@0:               'food (semisolid):FOODON_03430144',
cstrittmatter@0:               'food (sliced):FOODON_00002455',
cstrittmatter@0:               'food (smoked or smoke-flavored):FOODON_03460172',
cstrittmatter@0:               'food (solid):FOODON_03430151',
cstrittmatter@0:               'food (spoiled):FOODON_00003366',
cstrittmatter@0:               'food (starch or flour thickened):FOODON_03315268',
cstrittmatter@0:               'food (steamed):FOODON_00002657',
cstrittmatter@0:               'food (sugar-free):FOODON_03315838',
cstrittmatter@0:               'food (textured):FOODON_00002658',
cstrittmatter@0:               'food (toasted):FOODON_00002659',
cstrittmatter@0:               'food (unprocessed):FOODON_03316056',
cstrittmatter@0:               'food (unstandardized):FOODON_03315636',
cstrittmatter@0:              ]
cstrittmatter@0: 
cstrittmatter@0: fo_organism = ['algae:FOODON_03411301',
cstrittmatter@0:                'animal:FOODON_00003004',
cstrittmatter@0:                'fungus:FOODON_03411261',
cstrittmatter@0:                'lichen:FOODON_03412345',
cstrittmatter@0:                'whole plant:PO_0000003',
cstrittmatter@0:               ]
cstrittmatter@0: 
cstrittmatter@0: ncbi_taxon = ['Actinopterygii:NCBITaxon_7898', #mix of taxon types
cstrittmatter@0:               'Ecdysozoa:NCBITaxon_1206794',
cstrittmatter@0:               'Echinodermata:NCBITaxon_7586',
cstrittmatter@0:               'Fungi:NCBITaxon_4751',
cstrittmatter@0:               'Spiralia:NCBITaxon_2697495',
cstrittmatter@0:               'Viridiplantae:NCBITaxon_33090',
cstrittmatter@0:               'Amphibia:NCBITaxon_8292',
cstrittmatter@0:               #'Sauropsida:NCBITaxon_8457',
cstrittmatter@0:               'Aves:NCBITaxon_8782',
cstrittmatter@0:               'Crocodylia:NCBITaxon_1294634',
cstrittmatter@0:               'Testudinata:NCBITaxon_2841271',
cstrittmatter@0:               'Lepidosauria:NCBITaxon_8504',
cstrittmatter@0:               #'Mammalia:NCBITaxon_40674',
cstrittmatter@0:               'Artiodactyla:NCBITaxon_91561',
cstrittmatter@0:               'Carnivora:NCBITaxon_33554',
cstrittmatter@0:               'Chiroptera:NCBITaxon_9397',
cstrittmatter@0:               'Chrysochloridae:NCBITaxon_9389',
cstrittmatter@0:               'Eulipotyphla:NCBITaxon_9362',
cstrittmatter@0:               'Hyracoidea:NCBITaxon_9810',
cstrittmatter@0:               'Macroscelidea:NCBITaxon_28734',
cstrittmatter@0:               'Metatheria:NCBITaxon_9263',
cstrittmatter@0:               'Ornithorhynchidae:NCBITaxon_9256',
cstrittmatter@0:               'Perissodactyla:NCBITaxon_9787',
cstrittmatter@0:               'Pholidota:NCBITaxon_9971',
cstrittmatter@0:               'Primates:NCBITaxon_9443',
cstrittmatter@0:               'Proboscidea:NCBITaxon_9779',
cstrittmatter@0:               'Rodentia:NCBITaxon_9989',
cstrittmatter@0:               'Sirenia:NCBITaxon_9774',
cstrittmatter@0:               'Tachyglossidae:NCBITaxon_9259',
cstrittmatter@0:               'Tenrecidae:NCBITaxon_9369',
cstrittmatter@0:               'Tubulidentata:NCBITaxon_9815',
cstrittmatter@0:               'Xenarthra:NCBITaxon_9348',              
cstrittmatter@0:              ]
cstrittmatter@0: 
cstrittmatter@0: arg_bins = {#'fo_consumer':fo_consumer,
cstrittmatter@0:             'fo_product':fo_product,
cstrittmatter@0:             'fo_quality':fo_quality,
cstrittmatter@0:             'fo_organism':fo_organism,
cstrittmatter@0:             'ncbi_taxon':ncbi_taxon,
cstrittmatter@0:            }