Mercurial > repos > cstrittmatter > lexmapr2

annotate lexmapr/create_databases.py @ 0:f298f3e5c515

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload"
author cstrittmatter
date Mon, 27 Jun 2022 17:48:55 -0400
parents
children
rev   line source
cstrittmatter@0 1 """Builds SQLite3 databases"""
cstrittmatter@0 2
cstrittmatter@0 3 import logging, os, pickle, re, requests, sqlite3, sys, time
cstrittmatter@0 4 import lexmapr.ontology_reasoner as ontr
cstrittmatter@0 5 from nltk.tokenize import word_tokenize
cstrittmatter@0 6 from lexmapr.pipeline_helpers import punctuation_treatment
cstrittmatter@0 7 from lexmapr.definitions import embl_ontologies, synonym_db, ontol_db
cstrittmatter@0 8 from lexmapr.definitions import owl_dir, purl_link, missing_ontol_labels
cstrittmatter@0 9 from lexmapr.pipeline_resources import get_resource_label_permutations
cstrittmatter@0 10
cstrittmatter@0 11 logging.getLogger('requests').setLevel(logging.WARNING)
cstrittmatter@0 12 logging.getLogger('urllib3').setLevel(logging.WARNING)
cstrittmatter@0 13
cstrittmatter@0 14
cstrittmatter@0 15 # TODO: might replace pickle with ujson
cstrittmatter@0 16 def _pickle_save(data_to_save, file_path):
cstrittmatter@0 17 '''Write a pickle file'''
cstrittmatter@0 18 with open(file_path,'wb') as SAVE_file:
cstrittmatter@0 19 pickle.dump(data_to_save, SAVE_file)
cstrittmatter@0 20
cstrittmatter@0 21
cstrittmatter@0 22 def _pickle_load(file_path):
cstrittmatter@0 23 '''Read a pickle file'''
cstrittmatter@0 24 with open(file_path,'rb') as LOAD_file:
cstrittmatter@0 25 return(pickle.load(LOAD_file))
cstrittmatter@0 26
cstrittmatter@0 27
cstrittmatter@0 28 def _get_ontols(ontol_interest):
cstrittmatter@0 29 '''Obtain URLs for ontologies of interest'''
cstrittmatter@0 30 ontol_dic = {}
cstrittmatter@0 31 embl_resp = requests.get(embl_ontologies)
cstrittmatter@0 32 resp_blocks = re.findall('<tr>([\s\S]+?)</tr>',embl_resp.content.decode('utf-8'))
cstrittmatter@0 33 for resp_block in resp_blocks:
cstrittmatter@0 34 try:
cstrittmatter@0 35 embl_abbr = re.search('class=\"ontology-source\">([\s\S]+?)<', resp_block).group(1)
cstrittmatter@0 36 embl_name = re.search('<a href=\"/ols/ontologies/' + embl_abbr.lower() +\
cstrittmatter@0 37 '\">([\s\S]+?)</a>', resp_block).group(1)
cstrittmatter@0 38 embl_link = re.search('href=\"(\S+)\">Download', resp_block).group(1)
cstrittmatter@0 39 if embl_link.startswith('ontologies'):
cstrittmatter@0 40 embl_link = embl_link[len('ontologies'):]
cstrittmatter@0 41 # TODO: with Python 3.9- embl_link.removeprefix('ontologies')
cstrittmatter@0 42 except(AttributeError):
cstrittmatter@0 43 continue
cstrittmatter@0 44 if embl_abbr in ontol_interest:
cstrittmatter@0 45 ontol_dic[embl_abbr] = (embl_name, embl_link)
cstrittmatter@0 46 # Continue if not find all ontologies of interest specified in definitions.py
cstrittmatter@0 47 not_found = set(ontol_interest).difference(set(ontol_dic.keys()))
cstrittmatter@0 48 if not_found:
cstrittmatter@0 49 if len(not_found) == 1:
cstrittmatter@0 50 logging.warning(f'Did not find ontology: ' + ', '.join(not_found))
cstrittmatter@0 51 else:
cstrittmatter@0 52 logging.warning(f'Did not find ontologies: ' + ', '.join(not_found))
cstrittmatter@0 53 if ontol_dic == {}:
cstrittmatter@0 54 sys.exit('Zero ontologies found from user-given list')
cstrittmatter@0 55 return(ontol_dic)
cstrittmatter@0 56
cstrittmatter@0 57
cstrittmatter@0 58 def _check_make(db_file, remake_cache, ontol_interest):
cstrittmatter@0 59 '''Check if database file should be remade'''
cstrittmatter@0 60 if os.path.exists(db_file) and remake_cache == False:
cstrittmatter@0 61 if os.path.exists(os.path.join(owl_dir, 'cached_ontologies.pickle')):
cstrittmatter@0 62 if ontol_interest == _pickle_load(os.path.join(owl_dir, 'cached_ontologies.pickle')):
cstrittmatter@0 63 return(False)
cstrittmatter@0 64 try:
cstrittmatter@0 65 os.remove(db_file)
cstrittmatter@0 66 except(FileNotFoundError):
cstrittmatter@0 67 pass
cstrittmatter@0 68 return(True)
cstrittmatter@0 69
cstrittmatter@0 70
cstrittmatter@0 71 def _db_insert(db_cursor, table_name, key_term, val_term):
cstrittmatter@0 72 '''Insert new data into a database table'''
cstrittmatter@0 73 if key_term.strip()==val_term.strip() or key_term.strip()=='' or val_term.strip()=='':
cstrittmatter@0 74 return
cstrittmatter@0 75 db_cursor.execute(f"INSERT OR IGNORE INTO {table_name} VALUES (:key,:value)",
cstrittmatter@0 76 {'key':key_term.strip(), 'value':val_term.strip()})
cstrittmatter@0 77
cstrittmatter@0 78
cstrittmatter@0 79 def _get_imports(file_handle):
cstrittmatter@0 80 '''Check for required imports; append any new patterns to pattern_strs'''
cstrittmatter@0 81 pattern_strs = []
cstrittmatter@0 82 pattern_strs.append('<owl:imports rdf:resource=\"&obo;(.*)"/>')
cstrittmatter@0 83 pattern_strs.append('<owl:imports rdf:resource=\"(http://.*)\"/>')
cstrittmatter@0 84 whole_file = str(file_handle.read())
cstrittmatter@0 85 for patt_str in pattern_strs:
cstrittmatter@0 86 import_match = re.findall(patt_str, whole_file)
cstrittmatter@0 87 if import_match != []:
cstrittmatter@0 88 import_match = [x if re.search('^http:',x) else purl_link+x for x in import_match]
cstrittmatter@0 89 break
cstrittmatter@0 90 return(import_match)
cstrittmatter@0 91
cstrittmatter@0 92
cstrittmatter@0 93 def _section_file(file_handle, break_pattern,
cstrittmatter@0 94 stt_at=' // Classes', end_at=' // Annotations'):
cstrittmatter@0 95 '''Break OWL files into readable sections for each ontology accession'''
cstrittmatter@0 96 whole_file = str(file_handle.read())
cstrittmatter@0 97 if stt_at != '':
cstrittmatter@0 98 if re.search(stt_at, whole_file):
cstrittmatter@0 99 whole_file = ''.join(whole_file.split(stt_at)[1:])
cstrittmatter@0 100 if end_at != '':
cstrittmatter@0 101 if re.search(end_at, whole_file):
cstrittmatter@0 102 whole_file = ''.join(whole_file.split(end_at)[:-1])
cstrittmatter@0 103 file_sections = whole_file.split(break_pattern)
cstrittmatter@0 104 return(file_sections[1:-1])
cstrittmatter@0 105
cstrittmatter@0 106
cstrittmatter@0 107 def _labels_synonyms(obo_list, have_label=False):
cstrittmatter@0 108 '''Identify labels, ids and exact ontology synonyms'''
cstrittmatter@0 109 obo_ids = []
cstrittmatter@0 110 for obo_string in obo_list:
cstrittmatter@0 111 id_pattern = '(\w+) -->'
cstrittmatter@0 112 lab_pattern = '\<rdfs:label[\s\S]*?\>([\s\S]+?)\<\/rdfs:label\>'
cstrittmatter@0 113 syn_pattern = '\<oboInOwl:hasExactSynonym[\s\S]*?\>(.*?)\</oboInOwl:hasExactSynonym'
cstrittmatter@0 114 obo_id = re.search(id_pattern, obo_string).group(1)
cstrittmatter@0 115 # Do not save ids that are not formatted as expected, or placeholders/obsolete/Apollo:SV
cstrittmatter@0 116 if re.search('APOLLO:SV', obo_id) or re.search('APOLLO_SV', obo_id):
cstrittmatter@0 117 continue
cstrittmatter@0 118 elif obo_id in missing_ontol_labels:
cstrittmatter@0 119 continue
cstrittmatter@0 120 elif re.search(':', obo_id) or re.search('_', obo_id):
cstrittmatter@0 121 try:
cstrittmatter@0 122 obo_label = re.findall(lab_pattern, obo_string)[-1]
cstrittmatter@0 123 if re.search('^obsolete:', obo_label):
cstrittmatter@0 124 continue
cstrittmatter@0 125 except(IndexError):
cstrittmatter@0 126 obo_label = ''
cstrittmatter@0 127 obo_synonyms = re.findall(syn_pattern, obo_string)
cstrittmatter@0 128 obo_ids.append([str(obo_id.lstrip('_')), str(obo_label), obo_synonyms])
cstrittmatter@0 129 if have_label:
cstrittmatter@0 130 obo_ids = [x for x in obo_ids if x[1] != '']
cstrittmatter@0 131 return(obo_ids)
cstrittmatter@0 132
cstrittmatter@0 133
cstrittmatter@0 134 def get_synonyms(remake_cache, ontol_interest):
cstrittmatter@0 135 '''Create database of predefined synonyms'''
cstrittmatter@0 136 os.makedirs(owl_dir, exist_ok=True)
cstrittmatter@0 137 if _check_make(synonym_db, remake_cache, ontol_interest) == False:
cstrittmatter@0 138 return
cstrittmatter@0 139
cstrittmatter@0 140 conn = sqlite3.connect(synonym_db)
cstrittmatter@0 141 c = conn.cursor()
cstrittmatter@0 142 c.execute("CREATE TABLE label_synonyms (key TEXT, value TEXT, UNIQUE(key, value))")
cstrittmatter@0 143 print('{:<40}'.format('\tRetrieving predefined synonyms'),end='\r')
cstrittmatter@0 144 with open('lexmapr/predefined_resources/resource_synonyms.csv', 'r') as IN_file:
cstrittmatter@0 145 for read_line in IN_file.readlines():
cstrittmatter@0 146 split_line = read_line.split(',')
cstrittmatter@0 147 try:
cstrittmatter@0 148 ontol_name = split_line[2].split('_')[0].upper()
cstrittmatter@0 149 except(IndexError):
cstrittmatter@0 150 ontol_name = ontol_interest[0]
cstrittmatter@0 151 if ontol_name in ontol_interest:
cstrittmatter@0 152 _db_insert(c, 'label_synonyms', punctuation_treatment(split_line[0]),
cstrittmatter@0 153 punctuation_treatment(split_line[1]))
cstrittmatter@0 154 conn.commit()
cstrittmatter@0 155 with open('lexmapr/predefined_resources/SynLex.csv', 'r') as IN_file:
cstrittmatter@0 156 IN_file.readline()
cstrittmatter@0 157 for read_line in IN_file.readlines():
cstrittmatter@0 158 split_line = read_line.split(',')
cstrittmatter@0 159 try:
cstrittmatter@0 160 ontol_name = split_line[2].split('_')[0].upper().strip()
cstrittmatter@0 161 except(IndexError):
cstrittmatter@0 162 ontol_name = ontol_interest[0]
cstrittmatter@0 163 if ontol_name in ontol_interest:
cstrittmatter@0 164 _db_insert(c, 'label_synonyms', punctuation_treatment(split_line[0]),
cstrittmatter@0 165 punctuation_treatment(split_line[1]))
cstrittmatter@0 166 conn.commit()
cstrittmatter@0 167 conn.close()
cstrittmatter@0 168 return
cstrittmatter@0 169
cstrittmatter@0 170
cstrittmatter@0 171 def get_resource_ids(remake_cache, ontol_interest):
cstrittmatter@0 172 '''Create database of online resources and update synonym database'''
cstrittmatter@0 173 if _check_make(ontol_db, remake_cache, ontol_interest) == False:
cstrittmatter@0 174 return
cstrittmatter@0 175
cstrittmatter@0 176 conn = sqlite3.connect(ontol_db)
cstrittmatter@0 177 c = conn.cursor()
cstrittmatter@0 178 c.execute("CREATE TABLE non_standard_resource_ids (key TEXT, value TEXT, UNIQUE(key))")
cstrittmatter@0 179 c.execute("CREATE TABLE standard_resource_labels (key TEXT, value TEXT, UNIQUE(key, value))")
cstrittmatter@0 180 c.execute("""CREATE TABLE standard_resource_permutations (key TEXT, value TEXT,
cstrittmatter@0 181 UNIQUE(key, value))""")
cstrittmatter@0 182 sonn = sqlite3.connect(synonym_db)
cstrittmatter@0 183 s = sonn.cursor()
cstrittmatter@0 184
cstrittmatter@0 185 ontol_dic = _get_ontols(ontol_interest)
cstrittmatter@0 186 ontol_urls = [purl_link+x[1] for x in ontol_dic.values()]
cstrittmatter@0 187 for ontol_name in ontol_dic:
cstrittmatter@0 188 print('{:<40}'.format('\tRetrieving '+ontol_name[:15]+' terms'), end='\r')
cstrittmatter@0 189 no_label = set()
cstrittmatter@0 190 found_label = set()
cstrittmatter@0 191
cstrittmatter@0 192 if not os.path.isfile(os.path.join(owl_dir,ontol_name+'.owl')):
cstrittmatter@0 193 owl_download = requests.get(embl_ontologies+ontol_dic[ontol_name][1])
cstrittmatter@0 194 with open(os.path.join(owl_dir,ontol_name+'.owl'), 'w', encoding='utf-8') as O_file:
cstrittmatter@0 195 O_file.write(owl_download.content.decode('utf-8'))
cstrittmatter@0 196
cstrittmatter@0 197 with open(os.path.join(owl_dir,ontol_name+'.owl'),'r', encoding='utf-8') as OWL_file:
cstrittmatter@0 198 owl_sections = _section_file(OWL_file, '<!-- '+purl_link)
cstrittmatter@0 199 obo_ids = _labels_synonyms(owl_sections)
cstrittmatter@0 200 OWL_file.seek(0)
cstrittmatter@0 201 import_urls = [x for x in _get_imports(OWL_file) if x not in ontol_urls]
cstrittmatter@0 202
cstrittmatter@0 203 for imp_url in import_urls:
cstrittmatter@0 204 imp_download = requests.get(imp_url)
cstrittmatter@0 205 imp_file = os.path.join(owl_dir,imp_url.split('/')[-1])
cstrittmatter@0 206 with open(imp_file, 'w', encoding='utf-8') as T_file:
cstrittmatter@0 207 T_file.write(imp_download.content.decode('utf-8'))
cstrittmatter@0 208 # Assume possible to get duplicate names for different imports, so rewrite file
cstrittmatter@0 209 with open(imp_file, 'r', encoding='utf-8') as T_file:
cstrittmatter@0 210 owl_sections = _section_file(T_file, '<!-- '+purl_link, stt_at='', end_at='')
cstrittmatter@0 211 obo_ids.extend(_labels_synonyms(owl_sections, have_label=True))
cstrittmatter@0 212 os.remove(imp_file)
cstrittmatter@0 213 # TODO: more elegant way to find which ids don't have labels and to remove duplicates?
cstrittmatter@0 214 [no_label.add(x[0]) if x[1]=='' else found_label.add(x[0]) for x in obo_ids]
cstrittmatter@0 215 no_label_found = list(no_label.difference(found_label))
cstrittmatter@0 216 for ontol_term in obo_ids:
cstrittmatter@0 217 if ontol_term[0] in no_label_found:
cstrittmatter@0 218 ontol_acc = ontr.Ontology_accession(':'+ontol_term[0])
cstrittmatter@0 219 if ontol_acc.label != 'unk':
cstrittmatter@0 220 if not re.search('^obsolete:', ontol_acc.label):
cstrittmatter@0 221 ontol_term[1] = ontol_acc.label
cstrittmatter@0 222 time.sleep(0.05)
cstrittmatter@0 223 if ontol_term[1] == '':
cstrittmatter@0 224 continue
cstrittmatter@0 225 ontol_label = punctuation_treatment(ontol_term[1])
cstrittmatter@0 226 _db_insert(c, 'non_standard_resource_ids', ontol_term[0], ontol_term[1])
cstrittmatter@0 227 _db_insert(c, 'standard_resource_labels', ontol_label, ontol_term[0])
cstrittmatter@0 228 # Note from v 0.7: To limit performance overhead, we ignore resource labels with
cstrittmatter@0 229 # more than 7 tokens, as permutating too many tokens can be costly.
cstrittmatter@0 230 # Ignore NCBI taxon terms
cstrittmatter@0 231 if len(word_tokenize(ontol_label)) < 7 and len(word_tokenize(ontol_label)) > 1:
cstrittmatter@0 232 if not re.search('NCBITaxon', ontol_term[0]):
cstrittmatter@0 233 for permutation in get_resource_label_permutations(ontol_label):
cstrittmatter@0 234 _db_insert(c,'standard_resource_permutations',permutation,ontol_term[0])
cstrittmatter@0 235 # Add abbreviated binomials from NCBITaxons; TODO: may get wrong combinations?
cstrittmatter@0 236 elif len(word_tokenize(ontol_label)) == 2:
cstrittmatter@0 237 bi_name = ontol_label.split()
cstrittmatter@0 238 _db_insert(c, 'standard_resource_permutations',
cstrittmatter@0 239 bi_name[0][0]+' '+bi_name[1], ontol_term[0])
cstrittmatter@0 240 for syn_term in ontol_term[2]:
cstrittmatter@0 241 _db_insert(s,'label_synonyms',punctuation_treatment(str(syn_term)),ontol_label)
cstrittmatter@0 242 conn.commit()
cstrittmatter@0 243 sonn.commit()
cstrittmatter@0 244
cstrittmatter@0 245 conn.close()
cstrittmatter@0 246 sonn.close()
cstrittmatter@0 247 _pickle_save(ontol_interest, os.path.join(owl_dir,'cached_ontologies.pickle'))
cstrittmatter@0 248 return