view lexmapr/definitions.py @ 0:f5c39d0447be

"planemo upload"
author kkonganti
date Wed, 31 Aug 2022 14:32:07 -0400
parents
children
line wrap: on
line source
"""Static definitions"""

import os


# root path
ROOT = os.path.dirname(__file__)

# URL to list of OLS ontologies where download link is given
embl_ontologies = 'https://www.ebi.ac.uk/ols/ontologies'

# beginning of URL to ontology PURL
purl_link = 'http://purl.obolibrary.org/obo/'

# directory for downloaded OWL files
owl_dir = 'lexmapr/owl_files'

# path to database with synonyms from predefined resources and from OWL files
synonym_db = 'lexmapr/owl_files/label_synonyms.db'

# path to database with all ontologies of interest
ontol_db = 'lexmapr/owl_files/ontol_table.db'


# ontologies of interest
ontol_interest = [#'BFO',
                  #'CHEBI',
                  #'ENVO',
                  'FOODON',
                  #'GENEPIO',
                  'NCBITAXON', # NCBITaxon is not valid as of April 2022
                  #'OGMS',
                  #'PATO',
                  #'PCO',
                  #'UBERON',
                 ]

# ontology accessions that do not have labels or are placeholders as of April 2022
# will skip in database building
missing_ontol_labels = ['GENEPIO_0001367','GENEPIO_0001368','GENEPIO_0001369','GENEPIO_0001370',
                        'GENEPIO_0001372','GENEPIO_0001373','_MIAA_0000021',
                       ]

# terms indicating that the metadata was not given/collected; will output empty results
not_provided = ['not applicable','unknown','n a','not provided','not available','miscellaneous',
                'not collected','missing','unidentified','unknown','none','unamed','other',
                'undetermined','not known','no history given','no source specified','null',
                'unspecified','not reported','not available not collected','not isolated',
                'not available','not provided','xxx','mising','misng','other','unidentified',
                'not determined other','reported later','intact unknown','not determined',
                'not ascertained','unk','nd','nd others','nd other','etc','na','',' ',
                'not supplied','not specified',
               ]

# below are bin definitions
# TODO: food consumer group:FOODON_03510136 changed, formatted as '* as food consumer'
#       can collect as 'for *' in text?
#fo_consumer = []

fo_product = ['algal food product:FOODON_00001184',
              'amphibian:FOODON_03411624',
              'amphibian or reptile food product:FOODON_00002200',
              'animal based refined or partially-refined food product:FOODON_00001595',
              'avian egg food product:FOODON_00001105',
              'avian food product:FOODON_001251',
              'bakery food product:FOODON_00001626',
              'cell-based technology food product:FOODON_00003376',
              'dairy food product:FOODON_00001256',
              'dietary supplement:FOODON_03401298',
              'fish egg food product:FOODON_00001250',
              'fish food product:FOODON_00001248',
              'food product analog:FOODON_00001871',
              'food product component:FOODON_00001714',
              'fungus food product:FOODON_00001143',
              'game animal food product:FOODON_00002477',
              'insect food product:FOODON_00001177',
              'meat food product:FOODON_00002477',
              'microbial food product:FOODON_00001145',
              'plant food product:FOODON_00001015',
              'poultry food product:FOODON_00001283',
              'prepared food product:FOODON_00001180',
              'processed food product:FOODON_03311737',
              'reptile egg food product:FOODON_00002199',
              'seafood product:FOODON_00001046',
              'shellfish food product:FOODON_00001293',
              'soup food product:FOODON_00002257',
              'sustainable agriculture food product:FOODON_00003375',
              'vegetarian food product:FOODON_00003194',
              'vertebrate animal food product:FOODON_00001092',
             ]

fo_quality = ['food (acidified):FOODON_03301625',
              'food (adulterated):FOODON_00003367',
              'food (baked):FOODON_00002456',
              'food (batter-coated):FOODON_00002662',
              'food (blanched):FOODON_00002767',
              'food (blend):FOODON_00003889',
              'food (boiled):FOODON_00002688',
              'food (breaded):FOODON_00002661',
              'food (broiled or grilled):FOODON_00002647',
              'food (canned):FOODON_00002418',
              'food (chilled):FOODON_00002642',
              'food (chopped):FOODON_00002777',
              'food (cleaned):FOODON_00002708',
              'food (colored):FOODON_00002650',
              'food (comminuted):FOODON_00002754',
              'food (cooked):FOODON_00001181',
              'food (deep-fried):FOODON_03307052',
              'food (dehydrated):FOODON_00002643',
              'food (dried):FOODON_03307539',
              'food (fat or oil coated):FOODON_03460233',
              'food (fermented):FOODON_00001258',
              'food (filled):FOODON_00002644',
              'food (flavored):FOODON_00002646',
              'food (freeze-dried):FOODON_03301752',
              'food (fresh):FOODON_00002457',
              'food (fried):FOODON_00002660',
              'food (frozen):FOODON_03302148',
              'food (genetically-modified):FOODON_03530251',
              'food (ground):FOODON_00002713',
              'food (harvested):FOODON_00003398',
              'food (heat treated):FOODON_03316043',
              'food (hulled):FOODON_00002720',
              'food (hydrolized):FOODON_00002653',
              'food (irradiated):FOODON_03305364',
              'food (juiced):FOODON_00003499',
              'food (liquid):FOODON_03430130',
              'food (milled):FOODON_00002649',
              'food (not genetically-modified):FOODON_00003379',
              'food (organically grown):FOODON_03306690',
              'food (packaged):FOODON_00002739',
              'food (packed in high pressurised containers):FOODON_03317139',
              'food (pan-fried):FOODON_00002463',
              'food (paste):FOODON_00003887',
              'food (pasteurized):FOODON_00002654',
              'food (peeled):FOODON_00002655',
              'food (pickled):FOODON_00001079',
              'food (powdered):FOODON_00002976',
              'food (precooked):FOODON_00002971',
              'food (precooked, frozen):FOODON_03305323',
              'food (preserved):FOODON_00002158',
              'food (puffed):FOODON_00002656',
              'food (raw):FOODON_03311126',
              'food (rehydrated):FOODON_00002755',
              'food (roasted):FOODON_00002744',
              'food (salted):FOODON_03460173',
              'food (seasoned):FOODON_00002733',
              'fruit (seedless):FOODON_00003461',
              'food (semiliquid):FOODON_03430103',
              'food (semisolid):FOODON_03430144',
              'food (sliced):FOODON_00002455',
              'food (smoked or smoke-flavored):FOODON_03460172',
              'food (solid):FOODON_03430151',
              'food (spoiled):FOODON_00003366',
              'food (starch or flour thickened):FOODON_03315268',
              'food (steamed):FOODON_00002657',
              'food (sugar-free):FOODON_03315838',
              'food (textured):FOODON_00002658',
              'food (toasted):FOODON_00002659',
              'food (unprocessed):FOODON_03316056',
              'food (unstandardized):FOODON_03315636',
             ]

fo_organism = ['algae:FOODON_03411301',
               'animal:FOODON_00003004',
               'fungus:FOODON_03411261',
               'lichen:FOODON_03412345',
               'whole plant:PO_0000003',
              ]

ncbi_taxon = ['Actinopterygii:NCBITaxon_7898', #mix of taxon types
              'Ecdysozoa:NCBITaxon_1206794',
              'Echinodermata:NCBITaxon_7586',
              'Fungi:NCBITaxon_4751',
              'Spiralia:NCBITaxon_2697495',
              'Viridiplantae:NCBITaxon_33090',
              'Amphibia:NCBITaxon_8292',
              #'Sauropsida:NCBITaxon_8457',
              'Aves:NCBITaxon_8782',
              'Crocodylia:NCBITaxon_1294634',
              'Testudinata:NCBITaxon_2841271',
              'Lepidosauria:NCBITaxon_8504',
              #'Mammalia:NCBITaxon_40674',
              'Artiodactyla:NCBITaxon_91561',
              'Carnivora:NCBITaxon_33554',
              'Chiroptera:NCBITaxon_9397',
              'Chrysochloridae:NCBITaxon_9389',
              'Eulipotyphla:NCBITaxon_9362',
              'Hyracoidea:NCBITaxon_9810',
              'Macroscelidea:NCBITaxon_28734',
              'Metatheria:NCBITaxon_9263',
              'Ornithorhynchidae:NCBITaxon_9256',
              'Perissodactyla:NCBITaxon_9787',
              'Pholidota:NCBITaxon_9971',
              'Primates:NCBITaxon_9443',
              'Proboscidea:NCBITaxon_9779',
              'Rodentia:NCBITaxon_9989',
              'Sirenia:NCBITaxon_9774',
              'Tachyglossidae:NCBITaxon_9259',
              'Tenrecidae:NCBITaxon_9369',
              'Tubulidentata:NCBITaxon_9815',
              'Xenarthra:NCBITaxon_9348',              
             ]

arg_bins = {#'fo_consumer':fo_consumer,
            'fo_product':fo_product,
            'fo_quality':fo_quality,
            'fo_organism':fo_organism,
            'ncbi_taxon':ncbi_taxon,
           }