def match_numbering(ref_pdb, renum_pdb):
"""
Given two perfectly aligned pdbs (allowing offset in residue numbering),
match the numbering in renum_pdb to ref_pdb
It is definitely somewhat redundant to the:
def renumber_pdb( full_length_fasta_fn, truncated_pdb_fn, alignments=None ):
where the perfect function will be spliting sequences mapping (take the index mapping dictionary),
and renumbering as two separate functions
I did it for the quick and dirty trying not to mess up the working renumber_pdb() function
"""
ref_fasta = seq_util.pdb2fasta(ref_pdb)
renum_fasta = seq_util.pdb2fasta(renum_pdb)
if ref_fasta != renum_fasta and ref_fasta.index(renum_fasta):
offset = ref_fasta.index(renum_fasta)
idx2refpdbnum = idx_to_pdbnum(ref_pdb, offset)
pdb_as_ref = ref_fasta
pdb_to_renum = offset * "-" + renum_fasta
assert len(pdb_as_ref) == len(pdb_to_renum)
print "pdb_as_ref: %s" % pdb_as_ref
print "pdb_to_renum: %s" % pdb_to_renum
else:
assert ref_fasta == renum_fasta
idx2refpdbnum = idx_to_pdbnum(ref_pdb)
#
pdbline_dict = create_xyzDict_bychain(renum_pdb)[1]
chains = pdbline_dict.keys()
assert len(chains) == 1, "this script does not deal with pdbs containing multiple chains"
pdbline_dict = pdbline_dict[chains[0]]
res_nums = sorted(pdbline_dict.keys()) # pdbline_dict[ chain ].keys()
print res_nums
output_line = ""
for idx, rsn in enumerate(res_nums):
newrsn = idx2refpdbnum[idx]
# write to the outfile based on
for line in pdbline_dict[rsn].split("\n")[:-1]: # [:-1], because the last item in the list is ''
output_line += line[0:22] + "%4s" % newrsn + line[26:] + "\n"
output_fn = get_uniq_outpdbname(renum_pdb, "_matched")
output_buff = open(output_fn, "w")
output_buff.write("REMARK numberings are matched to %s\n" % ref_pdb)
output_buff.write(output_line)
output_buff.write("TER\n")
output_buff.close()
return output_fn
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
def renumber_pdb(full_length_fasta_fn, truncated_pdb_fn, alignments=None):
""" renumber the pdb based on the alignment """
""" future plan:
separate the renumber function into two functions
1. make a mapping function to return a mapping_Dict based on the alignments
2. renumber pdb based on the mapping_Dict
140914:
fixed a bug caused by an error introduced by dynamic programming,
where an residue name in the begining of a gap is the same as the end of the gap
KTTTTTKG <- query sequence
K------G <- mistaken alignment
------KG <- correct alignment
"""
""" the alignment should be in a list as two strings ["query_aln", "template_aln"] """
from sys import stderr
import os
if not alignments:
# get sequences
import seq_util
truncated_seq = seq_util.pdb2fasta(truncated_pdb_fn)
full_length_seq = seq_util.fasta_file_reader(full_length_fasta_fn)
if not truncated_seq.strip() or not full_length_seq.strip():
print "ERROR:"
exit()
# get alignment from biopython
from Bio import pairwise2
align_results = pairwise2.align.globalms(
full_length_seq, truncated_seq, 5, -5, -15, -0.5, penalize_end_gaps=False
)
if not align_results:
print full_length_seq
print ">", truncated_seq
exit()
alignments = align_results[0]
# print
# print "> Biopython pairwise2 globalms"
# print " match: 5, mismatch= -5"
# print " gap_open_penalty: -15, gap_extend_penalty: -0.5"
# print " penalize_end_gaps=False"
# from Bio.SubsMat.MatrixInfo import blosum62
# alignments = pairwise2.align.globalds( full_length_seq, truncated_seq, blosum62, -15, -0.1 )[0]
else:
print "alignments are given by the argument"
# create a map, based on the alignments
assert alignments
full_length_aln = alignments[0]
truncated_aln = alignments[1]
print "full_length_aln ", full_length_aln
print "truncated_aln ", truncated_aln
print
"""debug: print out the index in the alignments
count = 1
for f, t in zip( full_length_aln, truncated_aln ):
print "%3s %s %s" %( count, f, t )
count += 1 #"""
assert len(full_length_aln) == len(truncated_aln)
assert "-" not in full_length_aln, "this is a really dumb script, doesn't renum pdbs with aa unmatched"
# sequence mapping
index = 1
map_Dict = {}
full_length_rsn = 1
for str_idx in range(0, len(full_length_aln)):
""" in this loop, there is no way you should check this since compared to the partial_thread alignments, the full_length_aln is from fasta directly, there is no "-" at all here """
"""if full_length_aln.strip()[str_idx] == "-":
str_idx -= 1
full_length_rsn -= 1"""
if truncated_aln.strip()[str_idx] == "-":
full_length_rsn += 1
continue
"""debug
print "%3s %3s %s %s %3s" %(str_idx+1, full_length_rsn, full_length_aln.strip()[str_idx], truncated_aln.strip()[str_idx], index)#"""
if index not in map_Dict.keys():
map_Dict[index] = full_length_rsn
else:
from sys import stderr
stderr.write("ERROR: something is wrong when indexing the alignments mapping.\n")
return 0
# print index, dict[str(index)]
index += 1
full_length_rsn += 1
"""# debug
for i in range( 0, len( full_length_aln )):
print i, map_Dict[ i ]#"""
# renumber the pdb based on the alignment mapping
from os import popen, remove, system
"""because the mapping assumes the truncated_pdb_fn starts with residue number 1, we need to renumber the input pdb to start at one"""
# should probably get rid of this dependency in the future
temp_file = basename(truncated_pdb_fn.split(".pdb")[0]) + ".temp.pdb"
renumberPDBs_script = "/net/em-stor4/Volumes/data/wangyr/scripts/pdb_utils/renumberPDBs.pl"
assert os.path.exists(renumberPDBs_script), "Error: %s doesn't exist" % renumberPDBs_script
os.system("%s -pdbfile %s -res1 1 > %s" % (renumberPDBs_script, truncated_pdb_fn, temp_file))
xyz_dict, pdbline_dict, resname_dict = create_xyzDict_bychain(temp_file)
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())
output_line = ""
for idx, rsn in enumerate(res_nums):
try:
next_rsn = res_nums[idx + 1]
except:
pass # do nothing since next_rsn will be equal to rsn
newrsn = map_Dict[rsn]
next_newrsn = map_Dict[next_rsn]
# detect chain break from alignment
if (next_newrsn - newrsn) > 1:
dist = cal_dist(xyz_dict[next_rsn]["CA"], xyz_dict[rsn]["CA"]) # dist from old numbering
stderr.write("chainbreak (from alignment) at %s-%s with dist %.3f\n" % (newrsn, next_newrsn, dist))
# if no physically chainbreak detected, possibly could be error in dynamic programming, trying fixing it by looking for same residue name in next_newrsn-1, if so, overwrite the corrent numbering to write to the next_newrsn-1
if dist <= 4.5:
stderr.write(
"WARNING: no gap physically detected from %s-%s; caught an error in dynamic programming; correcting now...\n"
% (newrsn, next_newrsn)
)
# index in seq, thus -1
assert full_length_seq[newrsn - 1] == full_length_seq[next_newrsn - 2], (
"ERROR: failing to looking for same residue name as pos:%s in pos:%s of fasta file could not fix the error in dynamic programming\n"
% (newrsn, next_newrsn - 1)
)
newrsn = next_newrsn - 1
# write to the outfile based on
for line in pdbline_dict[rsn].split("\n")[:-1]: # [:-1], because the last item in the list is ''
output_line += line[0:22] + "%4s" % newrsn + line[26:] + "\n"
output_fn = basename(truncated_pdb_fn[:-4]) + "_renum.pdb"
output_buff = open(output_fn, "w")
output_buff.write("REMARK full_length_aln %s\n" % full_length_aln)
output_buff.write("REMARK truncated_aln %s\n" % truncated_aln)
output_buff.write(output_line)
output_buff.write("TER\n")
output_buff.close()
os.remove(temp_file)
return output_fn
trimmed_fasta = parser.add_mutually_exclusive_group()
trimmed_fasta.add_argument("--truncated_pdb")
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())