def find_mammals(cursor, trivial_name_list):
mammals = []
for trivial_name in trivial_name_list:
switch_to_db(cursor, get_compara_name(cursor))
tax_id = trivial2taxid(cursor, trivial_name)
parent_id = taxid2parentid(cursor, tax_id)
tax_id = parent_id
is_mammal = False
while tax_id:
qry = "select name_txt from names where tax_id= %d " % int(tax_id)
qry += " and name_class = 'scientific name'"
rows = search_db(cursor, qry)
if rows and rows[0][0]:
if 'mammal' in rows[0][0].lower():
is_mammal = True
break
elif 'vertebrat' in rows[0][0].lower():
# if the thing wasa mammal, we would have found it by now
is_mammal = False
break
parent_id = taxid2parentid(cursor, tax_id)
if parent_id and parent_id > 1:
tax_id = parent_id
else:
tax_id = None
if is_mammal:
mammals.append(trivial_name)
return mammals
def get_common_name(cursor, species):
switch_to_db(cursor, get_compara_name(cursor))
tax_id = species2taxid(cursor, species)
switch_to_db(cursor, get_ncbi_tax_name(cursor))
qry = "select name_txt from names where "
qry += "tax_id = %d and " % tax_id
qry += "name_class = 'genbank common name'"
rows = search_db(cursor, qry)
if rows:
if ('ERROR' in rows[0]):
search_db(cursor, qry, verbose=True)
return ""
else:
return rows[0][0]
return ""
def fill_db_names_table(cursor, ensembl_db_name):
compara_db_name = get_compara_name(cursor)
# how is this suppesd to work if they keep pulling name out of their arses like this
exceptions = {'cricetulus_griseus': 'cricetulus_griseus_crigri',
'heterocephalus_glaber': 'heterocephalus_glaber_male',
# some ornythologists are confused
'cyanoderma_ruficeps': 'stachyris_ruficeps'}
# I am not sure what use if any I have from haveing both male and female H.glaber
# male annotation is newer
for species_name, db_name in ensembl_db_name.items():
name = species_name if species_name not in exceptions else exceptions[species_name]
qry = f"select genome_db_id from {compara_db_name}.genome_db where name='{name}'"
genome_db_id = hard_landing_search(cursor, qry)[0][0]
print(genome_db_id, species_name, db_name)
fixed_fields={"genome_db_id":genome_db_id, "species_name":species_name, "db_name":ensembl_db_name[species_name]}
store_or_update(cursor, "db_names", fixed_fields=fixed_fields, update_fields={}, primary_key='genome_db_id')
return
def get_trivial_names(cursor, all_species, common_name):
tax_id = {}
db_name = get_compara_name (cursor)
if (not db_name):
print("compara db not found")
exit(1)
switch_to_db(cursor, db_name)
for species in all_species:
tax_id[species] = ncbi_species2taxid(cursor, species)
trivial_names = set_trivial_names()
for species in all_species:
trivial = trivial_names.get(species, "not found")
common = common_name.get(species, "not found")
if trivial=="not found":
if common == "not found":
trivial = species.lower().replace(" ", "_").replace("'", "_")
else:
trivial = common.replace("common ","").lower().replace(" ", "_").replace("'", "_")
trivial_names[species] = trivial
return trivial_names, tax_id
def feed_name_shorthands (cursor, all_species):
short = {}
short['ailuropoda_melanoleuca'] = 'AME'
short['anas_platyrhynchos'] = 'APL'
short['anolis_carolinensis'] = 'ACA'
short['astyanax_mexicanus'] = 'AMX'
short['bos_taurus'] = 'BTA'
short['callithrix_jacchus'] = 'CJA'
short['canis_familiaris'] = 'CAF'
short['cavia_porcellus'] = 'CPO'
short['choloepus_hoffmanni'] = 'CHO'
short['danio_rerio'] = 'DAR'
short['dasypus_novemcinctus'] = 'DNO'
short['dipodomys_ordii'] = 'DOR'
short['echinops_telfairi'] = 'ETE'
short['equus_caballus'] = 'ECA'
short['erinaceus_europaeus'] = 'EEU'
short['felis_catus'] = 'FCA'
short['ficedula_albicollis'] = 'FAL'
short['gadus_morhua'] = 'GMO'
short['gallus_gallus'] = 'GAL'
short['gasterosteus_aculeatus'] = 'GAC'
short['gorilla_gorilla'] = 'GGO'
short['homo_sapiens'] = ''
short['ictidomys_tridecemlineatus'] = 'STO'
short['latimeria_chalumnae'] = 'LAC'
short['lepisosteus_oculatus'] = 'LOC'
short['loxodonta_africana'] = 'LAF'
short['macaca_mulatta'] = 'MMU'
short['macropus_eugenii'] = 'MEU'
short['meleagris_gallopavo'] = 'MGA'
short['microcebus_murinus'] = 'MIC'
short['monodelphis_domestica'] = 'MOD'
short['mus_musculus'] = 'MUS'
short['mustela_putorius_furo'] = 'MPU'
short['myotis_lucifugus'] = 'MLU'
short['nomascus_leucogenys'] = 'NLE'
short['ochotona_princeps'] = 'OPR'
short['oreochromis_niloticus'] = 'ONI'
short['ornithorhynchus_anatinus'] = 'OAN'
short['oryctolagus_cuniculus'] = 'OCU'
short['oryzias_latipes'] = 'ORL'
short['ovis_aries'] = 'OAR'
short['otolemur_garnettii'] = 'OGA'
short['pan_troglodytes'] = 'PTR'
short['papio_anubis'] = 'PAN'
short['poecilia_formosa'] = 'PFO'
short['pelodiscus_sinensis'] = 'PSI'
short['petromyzon_marinus'] = 'PMA'
short['pongo_abelii'] = 'PPY'
short['procavia_capensis'] = 'PCA'
short['pteropus_vampyrus'] = 'PVA'
short['rattus_norvegicus'] = 'RNO'
short['sarcophilus_harrisii'] = 'SHA'
short['sorex_araneus'] = 'SAR'
short['sus_scrofa'] = 'SSC'
short['taeniopygia_guttata'] = 'TGU'
short['takifugu_rubripes'] = 'TRU'
short['tarsius_syrichta'] = 'TSY'
short['tetraodon_nigroviridis'] = 'TNI'
short['tupaia_belangeri'] = 'TBE'
short['tursiops_truncatus'] = 'TTR'
short['vicugna_pacos'] = 'VPA'
short['xenopus_tropicalis'] = 'XET'
short['xiphophorus_maculatus'] = 'XMA'
db_name = get_compara_name (cursor)
qry = "use %s " % db_name
search_db (cursor, qry)
table = 'species_name_shorthands'
# if the table does not exist, make it
if not check_table_exists (cursor, db_name, table):
qry = "CREATE TABLE " + table + " (id INT(10) PRIMARY KEY AUTO_INCREMENT)"
rows = search_db (cursor, qry)
if (rows): return False
qry = "ALTER TABLE %s ADD %s VARCHAR(100)" % (table, 'species')
rows = search_db (cursor, qry)
if (rows): return False
qry = "ALTER TABLE %s ADD %s VARCHAR(10)" % (table, 'shorthand')
rows = search_db (cursor, qry)
if (rows): return False
for species in all_species:
if short.has_key(species):
fixed_fields = {}
update_fields = {}
fixed_fields ['species'] = species
update_fields ['shorthand'] = short[species]
store_or_update (cursor, table, fixed_fields, update_fields)
else:
print "short for ", species, " not found "
short[species] = ""
def feed_trivial_names (cursor, all_species):
tax_id = {}
trivial = {}
trivial['ailuropoda_melanoleuca'] = 'panda'
trivial['anas_platyrhynchos'] = 'duck'
trivial['anolis_carolinensis'] = 'anole_lizard'
trivial['astyanax_mexicanus'] = 'blind_cavefish'
trivial['bos_taurus'] = 'cow'
trivial['callithrix_jacchus'] = 'marmoset'
trivial['canis_familiaris'] = 'dog'
trivial['cavia_porcellus'] = 'guinea_pig'
trivial['choloepus_hoffmanni'] = 'sloth'
trivial['danio_rerio'] = 'zebrafish'
trivial['dasypus_novemcinctus'] = 'armadillo'
trivial['dipodomys_ordii'] = 'kangaroo_rat'
trivial['echinops_telfairi'] = 'madagascar_hedgehog'
trivial['equus_caballus'] = 'horse'
trivial['erinaceus_europaeus'] = 'european_hedgehog'
trivial['felis_catus'] = 'cat'
trivial['ficedula_albicollis'] = 'flycatcher'
trivial['gadus_morhua'] = 'cod'
trivial['gallus_gallus'] = 'chicken'
trivial['gasterosteus_aculeatus'] = 'stickleback'
trivial['gorilla_gorilla'] = 'gorilla'
trivial['homo_sapiens'] = 'human'
trivial['ictidomys_tridecemlineatus'] = 'squirrel'
trivial['latimeria_chalumnae'] = 'coelacanth'
trivial['lepisosteus_oculatus'] = 'spotted_gar'
trivial['loxodonta_africana'] = 'elephant'
trivial['macaca_mulatta'] = 'macaque'
trivial['macropus_eugenii'] = 'wallaby'
trivial['meleagris_gallopavo'] = 'turkey'
trivial['microcebus_murinus'] = 'lemur'
trivial['monodelphis_domestica'] = 'opossum'
trivial['mus_musculus'] = 'mouse'
trivial['mustela_putorius_furo'] = 'ferret'
trivial['myotis_lucifugus'] = 'bat'
trivial['nomascus_leucogenys'] = 'gibbon'
trivial['ochotona_princeps'] = 'pika'
trivial['oreochromis_niloticus'] = 'tilapia'
trivial['ornithorhynchus_anatinus'] = 'platypus'
trivial['oryctolagus_cuniculus'] = 'rabbit'
trivial['oryzias_latipes'] = 'medaka'
trivial['otolemur_garnettii'] = 'galago_lemur'
trivial['ovis_aries'] = 'sheep'
trivial['pan_troglodytes'] = 'chimpanzee'
trivial['papio_anubis'] = 'baboon'
trivial['pelodiscus_sinensis'] = 'turtle'
trivial['petromyzon_marinus'] = 'lamprey'
trivial['poecilia_formosa'] = 'amazon_molly'
trivial['pongo_abelii'] = 'orangutan'
trivial['procavia_capensis'] = 'hyrax'
trivial['pteropus_vampyrus'] = 'flying_fox'
trivial['rattus_norvegicus'] = 'rat'
trivial['sarcophilus_harrisii'] = 'tasmanian_devil'
trivial['sorex_araneus'] = 'european_shrew'
trivial['sus_scrofa'] = 'pig'
trivial['taeniopygia_guttata'] = 'zebra_finch'
trivial['takifugu_rubripes'] = 'fugu'
trivial['tarsius_syrichta'] = 'tarsier'
trivial['tetraodon_nigroviridis'] = 'pufferfish'
trivial['tupaia_belangeri'] = 'tree_shrew'
trivial['tursiops_truncatus'] = 'dolphin'
trivial['vicugna_pacos'] = 'alpaca'
trivial['xenopus_tropicalis'] = 'xenopus'
trivial['xiphophorus_maculatus'] = 'platyfish'
db_name = get_compara_name (cursor)
if (not db_name):
print "compara db not found"
exit(1)
qry = "use %s " % db_name
search_db (cursor, qry)
for species in all_species:
tax_id[species] = species2taxid (cursor, species)
# switch to ncbi taxonomy database
db_name = get_ncbi_tax_name (cursor)
if (not db_name):
print "ncbi taxonomy db not found"
exit(1)
qry = "use %s " % db_name
search_db (cursor, qry)
for species in all_species:
if trivial.has_key(species):
fixed_fields = {}
update_fields = {}
fixed_fields ['tax_id'] = tax_id[species]
fixed_fields ['name_class'] = 'trivial'
update_fields['name_txt'] = trivial[species]
store_or_update (cursor, 'names', fixed_fields, update_fields)
else:
print "trivial for ", species, " not found "
trivial[species] = ""
return True
