Mercurial > repos > kkonganti > cfsan_lexmapr2
comparison lexmapr/definitions.py @ 0:f5c39d0447be
"planemo upload"
author | kkonganti |
---|---|
date | Wed, 31 Aug 2022 14:32:07 -0400 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:f5c39d0447be |
---|---|
1 """Static definitions""" | |
2 | |
3 import os | |
4 | |
5 | |
6 # root path | |
7 ROOT = os.path.dirname(__file__) | |
8 | |
9 # URL to list of OLS ontologies where download link is given | |
10 embl_ontologies = 'https://www.ebi.ac.uk/ols/ontologies' | |
11 | |
12 # beginning of URL to ontology PURL | |
13 purl_link = 'http://purl.obolibrary.org/obo/' | |
14 | |
15 # directory for downloaded OWL files | |
16 owl_dir = 'lexmapr/owl_files' | |
17 | |
18 # path to database with synonyms from predefined resources and from OWL files | |
19 synonym_db = 'lexmapr/owl_files/label_synonyms.db' | |
20 | |
21 # path to database with all ontologies of interest | |
22 ontol_db = 'lexmapr/owl_files/ontol_table.db' | |
23 | |
24 | |
25 # ontologies of interest | |
26 ontol_interest = [#'BFO', | |
27 #'CHEBI', | |
28 #'ENVO', | |
29 'FOODON', | |
30 #'GENEPIO', | |
31 'NCBITAXON', # NCBITaxon is not valid as of April 2022 | |
32 #'OGMS', | |
33 #'PATO', | |
34 #'PCO', | |
35 #'UBERON', | |
36 ] | |
37 | |
38 # ontology accessions that do not have labels or are placeholders as of April 2022 | |
39 # will skip in database building | |
40 missing_ontol_labels = ['GENEPIO_0001367','GENEPIO_0001368','GENEPIO_0001369','GENEPIO_0001370', | |
41 'GENEPIO_0001372','GENEPIO_0001373','_MIAA_0000021', | |
42 ] | |
43 | |
44 # terms indicating that the metadata was not given/collected; will output empty results | |
45 not_provided = ['not applicable','unknown','n a','not provided','not available','miscellaneous', | |
46 'not collected','missing','unidentified','unknown','none','unamed','other', | |
47 'undetermined','not known','no history given','no source specified','null', | |
48 'unspecified','not reported','not available not collected','not isolated', | |
49 'not available','not provided','xxx','mising','misng','other','unidentified', | |
50 'not determined other','reported later','intact unknown','not determined', | |
51 'not ascertained','unk','nd','nd others','nd other','etc','na','',' ', | |
52 'not supplied','not specified', | |
53 ] | |
54 | |
55 # below are bin definitions | |
56 # TODO: food consumer group:FOODON_03510136 changed, formatted as '* as food consumer' | |
57 # can collect as 'for *' in text? | |
58 #fo_consumer = [] | |
59 | |
60 fo_product = ['algal food product:FOODON_00001184', | |
61 'amphibian:FOODON_03411624', | |
62 'amphibian or reptile food product:FOODON_00002200', | |
63 'animal based refined or partially-refined food product:FOODON_00001595', | |
64 'avian egg food product:FOODON_00001105', | |
65 'avian food product:FOODON_001251', | |
66 'bakery food product:FOODON_00001626', | |
67 'cell-based technology food product:FOODON_00003376', | |
68 'dairy food product:FOODON_00001256', | |
69 'dietary supplement:FOODON_03401298', | |
70 'fish egg food product:FOODON_00001250', | |
71 'fish food product:FOODON_00001248', | |
72 'food product analog:FOODON_00001871', | |
73 'food product component:FOODON_00001714', | |
74 'fungus food product:FOODON_00001143', | |
75 'game animal food product:FOODON_00002477', | |
76 'insect food product:FOODON_00001177', | |
77 'meat food product:FOODON_00002477', | |
78 'microbial food product:FOODON_00001145', | |
79 'plant food product:FOODON_00001015', | |
80 'poultry food product:FOODON_00001283', | |
81 'prepared food product:FOODON_00001180', | |
82 'processed food product:FOODON_03311737', | |
83 'reptile egg food product:FOODON_00002199', | |
84 'seafood product:FOODON_00001046', | |
85 'shellfish food product:FOODON_00001293', | |
86 'soup food product:FOODON_00002257', | |
87 'sustainable agriculture food product:FOODON_00003375', | |
88 'vegetarian food product:FOODON_00003194', | |
89 'vertebrate animal food product:FOODON_00001092', | |
90 ] | |
91 | |
92 fo_quality = ['food (acidified):FOODON_03301625', | |
93 'food (adulterated):FOODON_00003367', | |
94 'food (baked):FOODON_00002456', | |
95 'food (batter-coated):FOODON_00002662', | |
96 'food (blanched):FOODON_00002767', | |
97 'food (blend):FOODON_00003889', | |
98 'food (boiled):FOODON_00002688', | |
99 'food (breaded):FOODON_00002661', | |
100 'food (broiled or grilled):FOODON_00002647', | |
101 'food (canned):FOODON_00002418', | |
102 'food (chilled):FOODON_00002642', | |
103 'food (chopped):FOODON_00002777', | |
104 'food (cleaned):FOODON_00002708', | |
105 'food (colored):FOODON_00002650', | |
106 'food (comminuted):FOODON_00002754', | |
107 'food (cooked):FOODON_00001181', | |
108 'food (deep-fried):FOODON_03307052', | |
109 'food (dehydrated):FOODON_00002643', | |
110 'food (dried):FOODON_03307539', | |
111 'food (fat or oil coated):FOODON_03460233', | |
112 'food (fermented):FOODON_00001258', | |
113 'food (filled):FOODON_00002644', | |
114 'food (flavored):FOODON_00002646', | |
115 'food (freeze-dried):FOODON_03301752', | |
116 'food (fresh):FOODON_00002457', | |
117 'food (fried):FOODON_00002660', | |
118 'food (frozen):FOODON_03302148', | |
119 'food (genetically-modified):FOODON_03530251', | |
120 'food (ground):FOODON_00002713', | |
121 'food (harvested):FOODON_00003398', | |
122 'food (heat treated):FOODON_03316043', | |
123 'food (hulled):FOODON_00002720', | |
124 'food (hydrolized):FOODON_00002653', | |
125 'food (irradiated):FOODON_03305364', | |
126 'food (juiced):FOODON_00003499', | |
127 'food (liquid):FOODON_03430130', | |
128 'food (milled):FOODON_00002649', | |
129 'food (not genetically-modified):FOODON_00003379', | |
130 'food (organically grown):FOODON_03306690', | |
131 'food (packaged):FOODON_00002739', | |
132 'food (packed in high pressurised containers):FOODON_03317139', | |
133 'food (pan-fried):FOODON_00002463', | |
134 'food (paste):FOODON_00003887', | |
135 'food (pasteurized):FOODON_00002654', | |
136 'food (peeled):FOODON_00002655', | |
137 'food (pickled):FOODON_00001079', | |
138 'food (powdered):FOODON_00002976', | |
139 'food (precooked):FOODON_00002971', | |
140 'food (precooked, frozen):FOODON_03305323', | |
141 'food (preserved):FOODON_00002158', | |
142 'food (puffed):FOODON_00002656', | |
143 'food (raw):FOODON_03311126', | |
144 'food (rehydrated):FOODON_00002755', | |
145 'food (roasted):FOODON_00002744', | |
146 'food (salted):FOODON_03460173', | |
147 'food (seasoned):FOODON_00002733', | |
148 'fruit (seedless):FOODON_00003461', | |
149 'food (semiliquid):FOODON_03430103', | |
150 'food (semisolid):FOODON_03430144', | |
151 'food (sliced):FOODON_00002455', | |
152 'food (smoked or smoke-flavored):FOODON_03460172', | |
153 'food (solid):FOODON_03430151', | |
154 'food (spoiled):FOODON_00003366', | |
155 'food (starch or flour thickened):FOODON_03315268', | |
156 'food (steamed):FOODON_00002657', | |
157 'food (sugar-free):FOODON_03315838', | |
158 'food (textured):FOODON_00002658', | |
159 'food (toasted):FOODON_00002659', | |
160 'food (unprocessed):FOODON_03316056', | |
161 'food (unstandardized):FOODON_03315636', | |
162 ] | |
163 | |
164 fo_organism = ['algae:FOODON_03411301', | |
165 'animal:FOODON_00003004', | |
166 'fungus:FOODON_03411261', | |
167 'lichen:FOODON_03412345', | |
168 'whole plant:PO_0000003', | |
169 ] | |
170 | |
171 ncbi_taxon = ['Actinopterygii:NCBITaxon_7898', #mix of taxon types | |
172 'Ecdysozoa:NCBITaxon_1206794', | |
173 'Echinodermata:NCBITaxon_7586', | |
174 'Fungi:NCBITaxon_4751', | |
175 'Spiralia:NCBITaxon_2697495', | |
176 'Viridiplantae:NCBITaxon_33090', | |
177 'Amphibia:NCBITaxon_8292', | |
178 #'Sauropsida:NCBITaxon_8457', | |
179 'Aves:NCBITaxon_8782', | |
180 'Crocodylia:NCBITaxon_1294634', | |
181 'Testudinata:NCBITaxon_2841271', | |
182 'Lepidosauria:NCBITaxon_8504', | |
183 #'Mammalia:NCBITaxon_40674', | |
184 'Artiodactyla:NCBITaxon_91561', | |
185 'Carnivora:NCBITaxon_33554', | |
186 'Chiroptera:NCBITaxon_9397', | |
187 'Chrysochloridae:NCBITaxon_9389', | |
188 'Eulipotyphla:NCBITaxon_9362', | |
189 'Hyracoidea:NCBITaxon_9810', | |
190 'Macroscelidea:NCBITaxon_28734', | |
191 'Metatheria:NCBITaxon_9263', | |
192 'Ornithorhynchidae:NCBITaxon_9256', | |
193 'Perissodactyla:NCBITaxon_9787', | |
194 'Pholidota:NCBITaxon_9971', | |
195 'Primates:NCBITaxon_9443', | |
196 'Proboscidea:NCBITaxon_9779', | |
197 'Rodentia:NCBITaxon_9989', | |
198 'Sirenia:NCBITaxon_9774', | |
199 'Tachyglossidae:NCBITaxon_9259', | |
200 'Tenrecidae:NCBITaxon_9369', | |
201 'Tubulidentata:NCBITaxon_9815', | |
202 'Xenarthra:NCBITaxon_9348', | |
203 ] | |
204 | |
205 arg_bins = {#'fo_consumer':fo_consumer, | |
206 'fo_product':fo_product, | |
207 'fo_quality':fo_quality, | |
208 'fo_organism':fo_organism, | |
209 'ncbi_taxon':ncbi_taxon, | |
210 } |