def extract_gene(uniformize_id):
gene_sequence = coding_gene_base[uniformize_id]
gene_sequence.id = get_id_protein(gene_sequence.id)
gene_sequence.name = ""
gene_sequence.description = ""
SeqIO.write(gene_sequence, gene_fasta_fl, "fasta")
protein_sequence = protein_base[uniformize_id]
protein_sequence.id = get_id_protein(protein_sequence.id)
protein_sequence.name = ""
protein_sequence.description = ""
SeqIO.write(protein_sequence, protein_fasta_fl, "fasta")
def one_head(df_row):
for R_cell in df_row:
if R_cell[0] == 'NA' or (str(R_cell[0]) == "NA"):
continue
body_list = R_cell[0].split()
if len(body_list) > 1:
one2two()
else:
uniformize_id = get_id_protein(body_list[0])
extract_gene(uniformize_id)
yield uniformize_id
def extract_gene(protein_file, gene_base_name, pan_gene_file_name,
gene_protein_mapping_table):
'''
input 1: pan protein fasta file
input 2: merged gene file
input 3: gene_protein_mapping_table
output 1:pan_gene_file_name
don't froget remove
'''
gene_base = SeqIO.index(
gene_base_name,
"fasta",
key_function=uniformize_gene_to_protein.get_id_gene)
with open(gene_protein_mapping_table, 'w+') as gene_protein_fl:
with open(pan_gene_file_name, 'w+') as out_fl:
for protein_sequence in SeqIO.parse(protein_file, "fasta"):
# if re.search("\.t1",protein_sequence.id) is not None: continue
# if re.search("^70-15",protein_sequence.id) is not None: continue
gene_id = uniformize_gene_to_protein.get_id_protein(
protein_sequence.id)
gene_sequence = gene_base[gene_id]
SeqIO.write(gene_sequence, out_fl, "fasta")
gene_protein_fl.write("{}\t{}\n".format(
gene_sequence.id, protein_sequence.id))
def extract_gene_two_head(protein_id):
gene_id = get_id_protein(protein_id)
gene_sequence = gene_base[gene_id]
return (gene_sequence)
def extract_gene(protein_id, out_fl):
gene_id = get_id_protein(protein_id)
gene_sequence = gene_base[gene_id]
SeqIO.write(gene_sequence, out_fl, "fasta")