def getFasta_proteic_DNA(self):
self.getFasta_seq_all_proteins()
self.getFasta_seq_proteins_DNA()
print("je suis diogo 2")
print(len(self.file_fasta_protein_seq))
print("je suis diogo")
print(type(self.file_fasta_protein_seq))
#print(self.file_fasta_protein_seq)
list_keys = []
list_proteins = []
for key, value in self.file_fasta_protein_seq.items():
list_keys.append(key)
DNA_string_key = self.get_DNA_description(key)
id_protein = -1
id_prot_DB_online = self.get_protein_id(
self.file_fasta_protein_seq[key].description)
designation = self.get_protein_description(
self.file_fasta_protein_seq[key].description)
sequence_prot = self.file_fasta_protein_seq[key].seq
sequence_DNA = self.file_fasta_nucleic_seq[DNA_string_key].seq
start_point, end_point = self.get_protein_location(
self.file_fasta_protein_seq[key].description)
protein_obj = Protein(id_protein, id_prot_DB_online, designation,
sequence_prot, sequence_DNA, start_point,
end_point)
print(start_point)
print(end_point)
list_proteins.append(protein_obj)
print(len(list_proteins))
print("finiinininininiiiiii")
return list_proteins
def createGeneProt(proteinOBJ, fkOrganism, fkContig = None):
"""
insert the bacterium, the whole genome, the proteins, and the genes
:param proteinOBJ: Protein object (old DB)
:param fkOrganism: id of the organism created
:param fkContig: id of the contig created in case of exists
:type proteinOBJ: Protein_new
:type fkOrganism: int
:type fkContig: int
"""
dna_head_gene = ''
if '>' not in proteinOBJ.designation:
dna_head_gene = '>' + proteinOBJ.designation
else:
dna_head_gene = proteinOBJ.designation
geneObjInsert = Gene(dna_head = dna_head_gene,
dna_sequence = proteinOBJ.sequence_dna, FK_id_organism = fkOrganism, start_position = proteinOBJ.start_point, end_position = proteinOBJ.end_point)
geneId = 0
if fkContig != None:
geneId = createGene(geneObjInsert, proteinOBJ.start_point_cnt, proteinOBJ.end_point_cnt, fkContig)
else:
geneId = createGene(geneObjInsert, None, None, None)
proteinOBJInsert = Protein(id_accession = proteinOBJ.id_accession, sequence_prot = proteinOBJ.sequence_prot, designation = proteinOBJ.designation)
if (proteinOBJ.sequence_prot != None):
proteinId = createProtein(proteinOBJInsert, fkOrganism, geneId)
def getProteinsByOrganism(idOrganism):
"""
get all the proteins of a given Organism
:param idOrganism: Id of the organism
:type idOrganism: int
:return: list of proteins
:rtype list[Protein]
"""
listProteins = []
qtyprots = 0
listProteins = Protein.get_all_Proteins_by_organism_id(idOrganism)
assert len(listProteins) > 1
writeCSVProteinNumber(idOrganism, len(listProteins))
return listProteins
contig_sequence_phage = ncbi_file_genbank.get_contig_of_the_phage()
list_ids_prot = ncbi_file_data_prot.get_list_protein_name()
list_of_proteins = []
for id_prot in list_ids_prot:
sequence_prot = ncbi_file_data_prot.get_protein_sequence_prot_id(
id_prot)
sequence_nuc = ncbi_file_data_nucleotid.get_nucleotid_sequence_prot_id(
id_prot)
vec_location = ncbi_file_data_nucleotid.get_location_by_prot_id(
id_prot)
protein_obj = Protein(id_accession=id_prot,
sequence_prot=sequence_prot,
sequence_dna=sequence_nuc,
start_point=vec_location[0],
end_point=vec_location[1])
list_of_proteins.append(protein_obj)
family_obj = None
genus_obj = None
specie_obj = None
strain_obj = None
family_obj = Family(designation='Phage no family')
genus_obj = Genus(designation='Phage no genuse')
specie_obj = Specie(designation='Phage no Specie')
strain_obj = Strain(designation=file_name[:-4])
print(strain_obj)
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