def align_and_renumber_pdb(fulllength_fasta, truncated_pdbfile, ignore_check=False):
""" This function is going to make the renumber_pdb() obsolete """
# make alignment
fl_seq = seq_util.fasta_file_reader(fulllength_fasta)
tc_seq = seq_util.pdb2fasta(truncated_pdbfile)
alignment = alignment_util.align_two_seqs(fl_seq, tc_seq)
if ignore_check:
pdb_idx1(
truncated_pdbfile, "temp.pdb"
) # for the following step, this has been used in alignment_util.correct_alignment_using_pdb
else:
alignment = alignment_util.correct_alignment_using_pdb(alignment, truncated_pdbfile, False)
seq_map = alignment_util.seq_mapping(alignment)
xyz_dict, pdbline_dict, resname_dict = create_xyzDict_bychain("temp.pdb")
assert len(pdbline_dict.keys()) == 1, (
"this script does not deal with pdbs containing multiple chains (%s)" % pdbline_dict.keys()
)
chain = pdbline_dict.keys()[0]
xyz_dict = xyz_dict[chain]
pdbline_dict = pdbline_dict[chain]
resname_dict = resname_dict[chain]
res_nums = sorted(pdbline_dict.keys())
out_pdblines = "REMARK full_length_aln %s\n" % alignment[0]
out_pdblines += "REMARK truncated_aln %s\n" % alignment[1]
for idx, rsn in enumerate(res_nums):
newrsn = seq_map[rsn]
for line in pdbline_dict[rsn].split("\n")[:-1]: # [:-1], because the last item in the list is ''
out_pdblines += line[0:22] + "%4s" % newrsn + line[26:] + "\n"
out_pdblines += "TER\n"
os.remove("temp.pdb")
return out_pdblines
trimmed_fasta.add_argument("--truncated_fasta")
parser.add_argument("--outfile_tag", default="trimmed", help="")
parser.add_argument("--debug", action="store_true", help="")
opts = parser.parse_args()
# read into fragment as frag[pos] = fragments
for fragfile in opts.fragfiles:
frag_dict = frag_util.read_fragfile(fragfile)
frag_len = frag_util.get_fraglen(frag_dict)
fl_seq = seq_util.fasta_file_reader(opts.fragfile_fasta)
if opts.truncated_pdb:
tc_seq = seq_util.pdb2fasta(opts.truncated_pdb)
alignment = alignment_util.correct_alignment_using_pdb(
alignment_util.align_two_seqs(fl_seq, tc_seq), opts.truncated_pdb
)
elif opts.truncated_fasta:
tc_seq = seq_util.fasta_file_reader(opts.truncated_fasta)
alignment = alignment_util.align_two_seqs(fl_seq, tc_seq)
else:
sys.stderr.write("ERROR: you need to either give --truncated_pdb or --truncated_fasta\n")
exit()
chainbreak_resnums = frag_util.get_positions_to_skip_from_alignment(alignment, frag_len)
seq_map = alignment_util.seq_mapping(alignment)
residues = sorted(seq_map.keys())
# residues before chain break shouldn't take