gen_bank_obj = Genbank_proteic_RAST(path_file_genBank)
    taxo = gen_bank_obj.get_taxonomy_array()
    print(gen_bank_obj.get_family())
    print(file_name)
    if len(taxo) == 7 or 'streptococcus_oralis' in file_name.lower():

        family_obj = None
        genus_obj = None
        specie_obj = None
        strain_obj = None
        if ('_phi' not in file_name.lower()
                or 'phage' not in file_name.lower()) and len(taxo) == 7:
            family_obj = Family(designation=gen_bank_obj.get_family())
            genus_obj = Genus(designation=gen_bank_obj.get_genus())
            specie_obj = Specie(designation=gen_bank_obj.get_specie())
            strain_obj = Strain(designation=gen_bank_obj.get_strain())
        if 'streptococcus_oralis' in file_name.lower():
            family_obj = Family(designation='Streptococcaceae')
            genus_obj = Genus(designation='Streptococcus')
            specie_obj = Specie(designation='Streptococcus Oralis')
            #name_bacterium_strain = file_name.split('-')[1]
            #name_bacterium_strain = file_name.split('-')[0] + "-" + name_bacterium_strain.split('_')[2]
            name_bacterium_strain = file_name
            strain_obj = Strain(designation=name_bacterium_strain)

        id_strain = strain_obj.get_id_strin_based_on_design_fk(6)
        print(id_strain)
        if id_strain == -1:
            family_obj.create_family()
            genus_obj.fk_family = family_obj.id_family
            genus_obj.create_genus()
Esempio n. 2
0
www = genbank_file.get_taxonomy_array()
print(www)

print(type(genbank_file.data_gen_bank))

print(list(genbank_file.data_gen_bank.keys())[0])

qty_contig = genbank_file.get_number_of_contigs()

print(genbank_file.get_family())

print(genbank_file.get_genus())

print(genbank_file.get_specie())

print(genbank_file.get_strain())

print("--------")



##### Start test for one organisme
start_time = time.time()
print("Start organism")
for contig in listas:
    print("pass contig: {0}".format(contig))
    list_prot = xls_obj.get_proteins_ids_by_contig_id(contig)
    for protein in list_prot:
        protein_info = xls_obj.get_information_line_by_protein_id(protein)

print("End organism")