def get_organism_type_by_strain(fkStrain):
"""
check the organism type and verify if any errors exists
:param fkStrain: FK strain of the organism
:type fkStrain: int
:return: error code or type of the organism (1 bacterium - 2 phage)
:rtype int
:note: -1 - more than one organism with this strain
-2 - No organism with this strain
-3 - Other errors
"""
listOrganism = Organism.get_organism_by_fk_strain(fkStrain)
qtyOrganism = len(listOrganism)
if qtyOrganism > 1:
return -1
elif qtyOrganism == 0:
return -2
elif qtyOrganism == 1:
return listOrganism[0].fk_type
else:
return -3
def get_bacterium_taxonomy_designation_by_id_bacterium(bacterium_id: int):
"""
return the strain, specie, and genus of the bacterium given it id
:return list with the respective taxonomy
:rtype list[str,str,str]
"""
organism_obj = Organism.get_organism_by_id(bacterium_id)
fk_strain = organism_obj.fk_strain
strain_obj = Strain.get_strain_by_id(fk_strain)
fk_specie = strain_obj.fk_specie
strain_designation = strain_obj.designation
specie_obj = Specie.get_specie_by_id(fk_specie)
fk_genus = specie_obj.fk_genus
specie_designation = specie_obj.designation
genus_id = specie_obj.fk_genus
genus_obj = Genus.get_genus_by_id(fk_genus)
genus_designation = genus_obj.designation
list_taxonomy = [strain_designation, specie_designation, genus_designation]
return list_taxonomy
def get_acc_by_organism(id_organism: int):
"""
return all the couple in the old db given the level id and ignore those obtain by Greg
:return List with the couples
:rtype List(Couple)
"""
organism_obj = Organism.get_organism_by_id(id_organism)
acc_number = organism_obj.acc_num
return acc_number
def getBacteriumStrainSpecieDesignationById(id_bacterium):
organism_obj = Organism.get_organism_by_id(id_bacterium)
id_strain = organism_obj.fk_strain
strain_obj = Strain.get_strain_by_id(id_strain)
id_specie = strain_obj.fk_specie
specie_obj = Specie.get_specie_by_id(id_specie)
specie_designation = specie_obj.designation
strain_designation = strain_obj.designation
taxonomy_bacterium = 'Specie: ' + specie_designation + ' Strain: ' + strain_designation
return taxonomy_bacterium
def get_phage_name_by_id(phage_id: int):
"""
return the name of the phage according to it id
:return phage name
:rtype str
"""
organism_obj = Organism.get_organism_by_id(phage_id)
fk_strain = organism_obj.fk_strain
strain_obj = Strain.get_strain_by_id(fk_strain)
strain_designation = strain_obj.designation
return strain_designation
print("---------------")
#print(line)
bact_id = line[1]['bact_ident']
bact_name = line[1]['bactphage']
p68_lysis = line[1]['P68']
p44AHJD_lysis = line[1]['44AHJD']
p3A_lysis = line[1]['3A']
p71_lysis = line[1]['71P']
p77_lysis = line[1]['77P']
pK_lysis = line[1]['K']
pSb1_lysis = line[1]['Sb-1']
print(bact_id)
print(bact_name)
print(p77_lysis)
print(pK_lysis)
dict_value = Organism.get_organism_id_by_strain_like(bact_name)
print(dict_value)
id_bact = bact_id
id_type_interaction = 1
type_interaction = True
id_level_interaction = 4
#def __init__(self, id_couple, interact_pn, fk_bacteria, fk_phage, fk_type_inter, fk_level_interact, fk_lysis_inter = -1, fk_source_data = -1):
#Check if any couple exists
id_phage = 10107
if p68_lysis != 'No':
print("3333333")
print(p68_lysis)
id_lysis = dict_lysis_type[p68_lysis]
def obtainphageACCnumberFromOldDBId(id_bacteriophage_old_db: int):
organism_obj = Organism.get_organism_by_id(id_bacteriophage_old_db)
acc_number = organism_obj.acc_num
return acc_number
def obtainBacteriumIdFromOldDBId(id_bacterium_old_db: int):
organism_obj = Organism.get_organism_by_id(id_bacterium_old_db)
fk_strain_old_db = organism_obj.fk_strain
return fk_strain_old_db
def getPhageDesignationById(id_phage):
organism_obj = Organism.get_organism_by_id(id_phage)
id_strain = organism_obj.fk_strain
strain_obj = Strain.get_strain_by_id(id_strain)
strain_designation = strain_obj.designation
return strain_designation
def load_get_bacteriophage(pathFile, arrayStates):
"""
read the CSV with the organism id, strain id and their states. This method call the other for the insertion of the bacteriophage on the database
:param pathFile: path of teh CSV file
:type pathFile: String
:return: matrix with all the data
:rtype panda dataframe
"""
arrayProblems = [11548, 11549]
wholeDNA = None
dataframe = pd.read_csv(pathFile)
for index, row in dataframe.iterrows():
if row['organism_type'] == 4:
row_verification = [[row['strain_db'], row['strain_api'], 2]]
id_strain_API = row['strain_api']
if id_strain_API not in arrayStates[:,
1] and id_strain_API not in arrayProblems:
print(row['strain_db'])
arrayStates = np.append(arrayStates, row_verification, axis=0)
writeCSVStateInsertion(arrayStates)
bacteriophageList = Organism.get_organism_by_fk_strain(
row['strain_db'])
assert len(bacteriophageList) == 1
#if bacteriophageList[0].qty_proteins > bacteriophageList[1].qty_proteins:
# bateriophage_take = 0
#else:
# bateriophage_take = 1
bateriophage_take = 0
bacteriophage = bacteriophageList[bateriophage_take]
organism_code = row['strain_db']
organism_codeAPI = row['strain_api']
accNumber = treatAccNumber(bacteriophage.acc_num,
organism_code)
bacteriophage.acc_num = accNumber
source_data = bacteriophage.fk_source_data
person_responsible = bacteriophage.fk_source
gi_number = treatGiNumber(bacteriophage.gi)
wholeDNAObj = getWholeGenomById(bacteriophage.fk_whole_genome)
strain = bacteriophage.fk_strain
print(strain)
listProts = getProteinsByOrganism(bacteriophage.id_organism)
listContigs = getContigsByOrganis(bacteriophage.id_organism)
if len(listContigs) > 0:
dictProts, dictConts = mapProteinsContigs(
listProts, listContigs)
canInsert, qtyProts = verifyQtyProtContigs(
len(listProts), dictProts, dictConts)
if canInsert == True:
print('Hello Contigs')
insertBacteriophageProteinsWholeDNAContigs(
bacteriophage, organism_code, wholeDNAObj,
dictProts, dictConts)
else:
writeCSVProteinContigNotEqual(
bacteriophage.id_organism, len(listProts),
qtyProts)
insertBacteriophageProteinsWholeDNA(
bacteriophage, organism_code, listProts,
wholeDNAObj)
else:
print("Hello")
insertBacteriophageProteinsWholeDNA(
bacteriophage, organism_code, listProts, wholeDNAObj)
row_verification = [row['strain_db'], row['strain_api'], 1]
arrayStates[-1] = row_verification
writeCSVStateInsertion(arrayStates)
def getProteinsDB(id_bacterium):
bacteriophage = Organism.get_organism_by_id(id_bacterium)
list_proteins = Protein.get_all_Proteins_by_organism_id(id_bacterium)
qty_prots_DB = len(list_proteins)
return qty_prots_DB
def load_get_bacterium(pathFile, arrayStates):
"""
read the CSV with the organism id, strain id and their states. This method call the other for the insertion of the bacterium on the database
:param pathFile: path of teh CSV file
:type pathFile: String
:return: matrix with all the data
:rtype panda dataframe
"""
wholeDNA = None
dataframe = pd.read_csv(pathFile)
for index, row in dataframe.iterrows():
print(row['organism_type'])
if row['organism_type'] == 1:
row_verification = [[row['strain_db'], row['strain_api'],2]]
id_strain_API = row['strain_api']
if id_strain_API not in arrayStates[:, 1]:
arrayStates = np.append(arrayStates, row_verification, axis=0)
writeCSVStateInsertion(arrayStates)
bacteriumList = Organism.get_organism_by_fk_strain(row['strain_db'])
assert len(bacteriumList) == 1
bacterium = bacteriumList[0]
bacterium.acc_num = bacterium.acc_num + '_V2'
#Test acc_value
bacterium_existance = checkBacteriumExistsByAcc(bacterium.acc_num)
#############
if bacterium_existance is True:
print('This acc {0} already exists'.format(bacterium.acc_num))
continue
accNumber = treatAccNumber(bacterium.acc_num)
source_data = bacterium.fk_source_data
person_responsible = bacterium.fk_source
gi_number = treatGiNumber(bacterium.gi)
wholeDNAObj = getWholeGenomById(bacterium.fk_whole_genome)
strain = bacterium.fk_strain
organism_code = row['strain_db']
organism_codeAPI = row['strain_api']
listProts = getProteinsByOrganism(bacterium.id_organism)
listContigs = getContigsByOrganis(bacterium.id_organism)
if len(listContigs)> 0 and len(listContigs[0].sequence) > 10:
dictProts, dictConts = mapProteinsContigs(listProts, listContigs)
canInsert, qtyProts = verifyQtyProtContigs(len(listProts),dictProts,dictConts)
if canInsert == True:
print('Hello)')
insertBacteriumProteinsWholeDNAContigs(bacterium, organism_codeAPI, wholeDNAObj, dictProts, dictConts)
elif qtyProts == 0:
insertBacteriumProteinsWholeDNAContigsNopositions(bacterium, organism_codeAPI, wholeDNAObj, dictProts, dictConts)
else:
writeCSVProteinContigNotEqual(bacterium.id_organism, len(listProts),qtyProts)
else:
print("Hello")
insertBacteriumProteinsWholeDNA(bacterium, organism_codeAPI, listProts, wholeDNAObj)
row_verification = [row['strain_db'], row['strain_api'],1]
arrayStates[-1] = row_verification
writeCSVStateInsertion(arrayStates)