def get_TRBC(FASTA_DIR):
# get the sequence representation for each TRA and TRB gene segment
# I'll just select the first allele for each sequence first
trbc_fasta = "/".join([FASTA_DIR, "TRBC.fasta"])
trb_c = Fasta.Fasta(trbc_fasta)
trb_c_dict = {}
trb_c_set = set()
trb_count = 0
for _fas in trb_c:
# parse the header
# select BL6
parse_head = _fas.header.split("|")
trb_gene = parse_head[1].split("*")[0]
trb_pos = parse_head[5]
if trb_gene not in trb_c_set:
trb_c_set.add(trb_gene)
trb_c_dict[trb_gene] = {
"Sequence": _fas.sequence,
"Gene": trb_gene,
"Length": _fas.length
}
trb_count += 1
return (trb_c_dict)
def create_dataset(n_success, fasta_dir, outfile):
'''
Create a data set of `n_success` valid TCRs, record valid and failed chains
'''
valid_counter = 0
alpha_v = []
beta_v = []
alpha_len = []
beta_len = []
alpha_diversity = []
beta_diversity = []
alpha_j = []
beta_j = []
alpha_aa = []
beta_aa = []
alpha_pep = []
beta_pep = []
alpha_keep = []
beta_keep = []
alpha_cdr1 = []
alpha_cdr2 = []
alpha_cdr3 = []
beta_cdr1 = []
beta_cdr2 = []
beta_cdr3 = []
alpha_cdr1_diversity = []
alpha_cdr2_diversity = []
alpha_cdr3_diversity = []
beta_cdr1_diversity = []
beta_cdr2_diversity = []
beta_cdr3_diversity = []
valid_tcr_chains = []
with open(outfile, "wt") as ofile:
ofile.write(
"TRA_V\tTRA_J\tTRA_len\tTRA_entropy\tTRA_AA\tTRA_Peptide\tTRA_Valid\tTRA_CDR1_entropy\tTRA_CDR1\tTRA_CDR2_entropy\tTRA_CDR2\tTRA_CDR3_entropy\tTRA_CDR3\tTRB_V\tTRB_J\tTRB_len\tTRB_entropy\tTRB_AA\tTRB_Peptide\tTRB_Valid\tTRB_CDR1_entropy\tTRB_CDR1\tTRB_CDR2_entropy\tTRB_CDR2\tTRB_CDR3_entropy\tTRB_CDR3\tValid.TCR\n"
)
while valid_counter < n_success:
tcr_chains = generate_TCR(1, fasta_dir)
for x in tcr_chains:
tcra = x['TCRA']
tcrb = x['TCRB']
alpha_v.append(tcra['V'])
alpha_j.append(tcra['J'])
alpha_len.append(len(tcra['RNA']))
# the entropy calculation should be for the CDR3 region or the V(D)J region only
# it should also be the translated AA sequence, not the nucleotide
alpha_rna_record = Fasta.FastaRecordPeptide(
header='', sequence=tcra['AA'])
alpha_n_count = pd.Series(alpha_rna_record.counts)
alpha_diversity.append(
stats.entropy(alpha_n_count / alpha_n_count.sum()))
alpha_aa.append(len(tcra['AA']))
alpha_pep.append(tcra['AA'])
alpha_keep.append(tcra["Valid.Chain"])
try:
# find the valid CDR regions if possible
# calculate the CDR entropies
alpha_cdr1_aa = get_CDR(tcra['AA'], 1)
alpha_aa_cdr1 = Fasta.FastaRecordPeptide(
header='', sequence=alpha_cdr1_aa)
alpha_cdr1_count = pd.Series(alpha_aa_cdr1.counts)
alpha_cdr1_en = stats.entropy(alpha_cdr1_count /
alpha_cdr1_count.sum())
alpha_cdr1_diversity.append(alpha_cdr1_en)
alpha_cdr2_aa = get_CDR(tcra['AA'], 2)
alpha_aa_cdr2 = Fasta.FastaRecordPeptide(
header='', sequence=alpha_cdr2_aa)
alpha_cdr2_count = pd.Series(alpha_aa_cdr2.counts)
alpha_cdr2_en = stats.entropy(alpha_cdr2_count /
alpha_cdr2_count.sum())
alpha_cdr2_diversity.append(alpha_cdr2_en)
try:
alpha_cdr3_aa = get_CDR(tcra['AA'], 3)
alpha_aa_cdr3 = Fasta.FastaRecordPeptide(
header='', sequence=alpha_cdr3_aa)
alpha_cdr3_count = pd.Series(alpha_aa_cdr3.counts)
alpha_cdr3_en = stats.entropy(alpha_cdr3_count /
alpha_cdr3_count.sum())
alpha_cdr3_diversity.append(alpha_cdr3_en)
except UnboundLocalError:
alpha_cdr3_aa = ""
alpha_cdr3_en = np.nan
alpha_cdr3_diversity.append(alpha_cdr3_en)
except TypeError:
alpha_cdr1_en = np.nan
alpha_cdr2_en = np.nan
alpha_cdr3_en = np.nan
alpha_cdr1_diversity.append(alpha_cdr1_en)
alpha_cdr2_diversity.append(alpha_cdr2_en)
alpha_cdr3_diversity.append(alpha_cdr3_en)
alpha_cdr1.append(alpha_cdr1_aa)
alpha_cdr2.append(alpha_cdr2_aa)
alpha_cdr3.append(alpha_cdr3_aa)
beta_v.append(tcrb['V'])
beta_j.append(tcrb['J'])
beta_len.append(len(tcrb['RNA']))
beta_rna_record = Fasta.FastaRecordPeptide(header='',
sequence=tcrb['AA'])
beta_n_count = pd.Series(beta_rna_record.counts)
beta_diversity.append(
stats.entropy(beta_n_count / beta_n_count.sum()))
beta_aa.append(len(tcrb['AA']))
beta_pep.append(tcrb['AA'])
beta_keep.append(tcrb["Valid.Chain"])
try:
# calculate the CDR entropies
beta_cdr1_aa = get_CDR(tcrb['AA'], 1)
beta_aa_cdr1 = Fasta.FastaRecordPeptide(
header='', sequence=beta_cdr1_aa)
beta_cdr1_count = pd.Series(beta_aa_cdr1.counts)
beta_cdr1_en = stats.entropy(beta_cdr1_count /
beta_cdr1_count.sum())
beta_cdr1_diversity.append(beta_cdr1_en)
beta_cdr2_aa = get_CDR(tcrb['AA'], 2)
beta_aa_cdr2 = Fasta.FastaRecordPeptide(
header='', sequence=beta_cdr2_aa)
beta_cdr2_count = pd.Series(beta_aa_cdr2.counts)
beta_cdr2_en = stats.entropy(beta_cdr2_count /
beta_cdr2_count.sum())
beta_cdr2_diversity.append(beta_cdr2_en)
try:
beta_cdr3_aa = get_CDR(tcrb['AA'], 3)
beta_aa_cdr3 = Fasta.FastaRecordPeptide(
header='', sequence=beta_cdr3_aa)
beta_cdr3_count = pd.Series(beta_aa_cdr3.counts)
beta_cdr3_en = stats.entropy(beta_cdr3_count /
beta_cdr3_count.sum())
beta_cdr3_diversity.append(beta_cdr3_en)
except UnboundLocalError:
beta_cdr3_aa = ""
beta_cdr3_en = np.nan
beta_cdr3_diversity.append(beta_cdr3_en)
except TypeError:
beta_cdr1_en = np.nan
beta_cdr2_en = np.nan
beta_cdr3_en = np.nan
beta_cdr1_diversity.append(beta_cdr1_en)
beta_cdr2_diversity.append(beta_cdr2_en)
beta_cdr3_diversity.append(beta_cdr3_en)
beta_cdr1.append(beta_cdr1_aa)
beta_cdr2.append(beta_cdr2_aa)
beta_cdr3.append(beta_cdr3_aa)
# check for validity
if tcra["Valid.Chain"] and tcrb["Valid.Chain"]:
valid_counter += 1
validchain = True
valid_tcr_chains.append(True)
else:
validchain = False
valid_tcr_chains.append(False)
ofile.write(
"{}\t{}\t{}\t{}\t{}\t\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\t{}\n"
.format(tcra['V'], tcra['J'], len(tcra['RNA']),
tcra['RNA'],
stats.entropy(alpha_n_count / alpha_n_count.sum()),
len(tcra['AA']), tcra['AA'], tcra["Valid.Chain"],
alpha_cdr1_en, alpha_cdr1_aa, alpha_cdr2_en,
alpha_cdr2_aa, alpha_cdr3_en,
alpha_cdr3_aa, tcrb['V'], tcrb['J'],
len(tcrb['RNA']), tcrb['RNA'],
stats.entropy(beta_n_count / beta_n_count.sum()),
len(tcrb['AA']), tcrb['AA'], tcrb["Valid.Chain"],
beta_cdr1_en, beta_cdr1_aa, beta_cdr2_en,
beta_cdr2_aa, beta_cdr3_en, beta_cdr3_aa,
validchain))