def find_min_vdel(v_nt, cdr3_aa):
# find the minimum number of V nucleotides need to be deleted to
# result in the CDR3aa
v_aa = lcommon.nt2aa(v_nt)
matchseq = left_max_match(v_aa, cdr3_aa) # max matched aa seq
len_ntmatch = len(matchseq) * 3
min_vdel = len(v_nt) - len_ntmatch
# see if 1 or 2 of the next v nts can result in the next cdr3 aa
if min_vdel > 0 and len(cdr3_aa) > len(matchseq):
if min_vdel >= 2:
v_nts = v_nt[len_ntmatch: len_ntmatch + 2]
else:
v_nts = v_nt[len_ntmatch]
got2nts = False
firstaa = cdr3_aa[len(matchseq)]
codons = lcommon.aa2codons(firstaa)
for codon in codons:
if re.match(v_nts, codon):
min_vdel -= len(v_nts)
got2nts = True
break
if len(v_nts) == 2 and not got2nts:
for codon in codons:
if re.match(v_nts[0], codon):
min_vdel -= 1
break
return min_vdel
def find_min_jdel(j_nt, cdr3_aa):
# find the minimum number of J nucleotides need to be deleted to
# result in the CDR3aa
hang_j = len(j_nt) % 3
hang_j_nts = j_nt[: hang_j] # the 5' of j nt seq that not part of a codon
j_aa = lcommon.nt2aa(j_nt[hang_j:])
matchseq = right_max_match(j_aa, cdr3_aa)
len_ntmatch = len(matchseq) * 3
min_jdel = len(j_nt) - len_ntmatch
if min_jdel > 0 and len(cdr3_aa) > len(matchseq):
if min_jdel >= 2:
j_nts = j_nt[min_jdel - 2: min_jdel]
else:
j_nts = j_nt[min_jdel - 1]
got2nts = False
firstaa = cdr3_aa[len(cdr3_aa) - len(matchseq) - 1]
codons = lcommon.aa2codons(firstaa)
for codon in codons:
if codon[-1 * len(j_nts): ] == j_nts:
min_jdel -= len(j_nts)
got2nts = True
break
if len(j_nts) == 2 and not got2nts:
for codon in codons:
if codon[2] == j_nts[1]:
min_jdel -= 1
break
return min_jdel
def get_all_codons(aaseq):
codon_lists = []
for aa in aaseq:
codon_lists.append(lcommon.aa2codons(aa))
return codon_lists
