def main():
#Initialize
options = parse_user_arguments()
verbose = options.show
pdb_path = os.path.join(config.get('Paths', 'modppi_path'),
config.get('Paths', 'pdb_path'))
dummy_dir = options.dummy_dir
try:
did_path = os.path.join(config.get('Paths', 'modppi_path'),
config.get('Paths', '3did_path'))
data_path = os.path.join(config.get('Paths', 'modppi_path'),
config.get('Paths', 'data_path'))
except:
did_path = options.outdir
data_path = options.outdir
if not os.path.exists(did_path):
os.makedirs(did_path)
if not os.path.exists(dummy_dir):
os.makedirs(dummy_dir)
if not os.path.exists(data_path):
sys.stderr.write(
"No DATA directory, please check your installation or INPUT\n")
#Parse did flat file
did = parse_3did(options)
#Create PDB files of 3DiD interactions
for dd, cases in did.iteritems():
for label in xrange(0, len(cases)):
#Define the name of the PDB output file with domain-domain interactions
did_file = os.path.join(
did_path, dd[0] + ":" + dd[1] + "#" + str(label) + ".brk.gz")
if not os.path.exists(did_file.lower()):
did_file = os.path.join(
did_path, dd[0] + ":" + dd[1] + "#" + str(label) + ".brk")
if not os.path.exists(did_file.lower()):
if verbose:
sys.stderr.write("\t\t--Create %s\n" % (did_file.lower()))
pdb_code, d1, d2 = cases[label]
pdb_file = os.path.join(pdb_path, pdb_code[1:3].lower(),
"pdb" + pdb_code + ".ent")
if not os.path.exists(pdb_file):
pdb_file = os.path.join(pdb_path, pdb_code[1:3].lower(),
"pdb" + pdb_code + ".ent.gz")
if not os.path.exists(pdb_file):
if verbose:
sys.stderr.write("\t\t\t-- %s not found\n" % pdb_file)
continue
try:
pdb = PDB(pdb_file)
brk = PDB()
pdb_chain_A = pdb.get_chain_by_id(d1[0])
start_A = d1[1]
end_A = d1[2]
pdb_chain_B = pdb.get_chain_by_id(d2[0])
start_B = d2[1]
end_B = d2[2]
brk_chain_A = pdb_chain_A.extract(init=start_A, end=end_A)
brk_chain_A.chain = "A"
brk.add_chain(brk_chain_A)
brk_chain_B = pdb_chain_B.extract(init=start_B, end=end_B)
brk_chain_B.chain = "B"
brk.add_chain(brk_chain_B)
brk.clean()
brk.write(did_file.lower())
except Exception as e:
if verbose: sys.stderr.write("\t\t\t Error: %s\n" % e)
continue
#Create list of interactions and FASTA sequences of 3DiD
did_interactions = open(os.path.join(data_path, options.interactions_file),
"w")
did_fasta = open(os.path.join(data_path, options.seq_file), "w")
for brk in os.listdir(did_path):
if verbose:
sys.stderr.write("\t\t-- Reading %s \n" %
os.path.join(did_path, brk))
try:
pdb = PDB(os.path.join(did_path, brk))
id_chain = []
for c in pdb.chain_identifiers:
pdb_chain = pdb.get_chain_by_id(c)
id_chain.append(pdb.id + "_" + c)
printfasta(did_fasta, pdb.id + "_" + c,
pdb_chain.gapped_protein_sequence)
did_interactions.write("%s\t%s\n" % (id_chain[0], id_chain[1]))
except Exception as e:
if verbose:
sys.stderr.write(
"\t\t-- %s cannot be read\n\t\t Error: %s\n" %
(os.path.join(did_path, brk), e))
continue
did_interactions.close()
did_fasta.close()
def renumber_pdb(config, path, pdb_name, sequences, dummy_dir):
'''
Renumber PDB file located in path folder with the real sequences
path Folder where PDB file is located
pdb PDB file
sequences dictionary of sequences (of ProteinSequence Class from SeqIO) that define the Aa number
chain identifier is the key of the dictionary
dummy_dir Dummy directory to cerate files
'''
#Initialize
from SBI.structure.chain import Chain
from SBI.sequence import Sequence
from SBI.structure import PDB
from Bio import SeqIO
from Bio import ExPASy
from Bio import AlignIO
from Bio.Align import Applications
clustal_exe = os.path.join(config.get('Paths', 'clustal_path'),
'clustalw2')
name_pdb = ".".join(pdb_name.split('/')[-1].split('.')[:-1])
new_pdb = PDB()
pdb_file = os.path.join(path, pdb_name)
pdb = PDB(pdb_file)
pdb.clean()
for chain_id, chain_seq in sequences.iteritems():
name_chain = name_pdb + "_" + chain_id
name_seq = chain_seq.get_identifier()
pdb_chain = pdb.get_chain_by_id(chain_id)
new_chain = Chain(name_pdb, chain_id)
#define/create files
infile = dummy_dir + "/tmp_" + name_chain + "_" + name_seq + ".fa"
outfile = dummy_dir + "/tmp_" + name_chain + "_" + name_seq + ".aln"
dndfile = dummy_dir + "/tmp_" + name_chain + "_" + name_seq + ".dnd"
fd = open(infile, "w")
fd.write(">{0:s}\n{1:s}\n".format(name_chain,
pdb_chain.protein_sequence))
fd.write(">{0:s}\n{1:s}\n".format(name_seq, chain_seq.get_sequence()))
fd.close()
try:
# run clustalw2
msa_cline = Applications.ClustalwCommandline(clustal_exe,
infile=infile,
outfile=outfile)
child = subprocess.Popen(str(msa_cline),
stdout=subprocess.PIPE,
stderr=subprocess.PIPE,
shell="/bin/bash")
child.communicate()
#store alignment in compare
alignment = AlignIO.read(outfile, 'clustal')
structure = alignment[0].seq
reference = alignment[1].seq
try:
len_3d = len(structure)
len_ref = len(reference)
except Exception as e:
sys.stderr.write("ERROR: %s\n" % e)
return e
except Exception as e:
sys.stderr.write("ERROR: %s\n" % e)
return e
#remove temporary fasta and alignment files
remove_files([infile, outfile, dndfile])
#mapping of residues to the original sequence
mapping = create_mapping(pdb_chain.protein_idx.split(";"), structure,
reference)
#fill the new chain with the correct numbering of residues
for residue in pdb_chain.aminoacids:
pair = (str(residue.number), residue.version)
number, version = mapping.get(pair)
residue.number = number
residue.version = version
new_chain.add_residue(residue)
#fill the new pdb
new_pdb.add_chain(new_chain)
return new_pdb
def modelling(queriesA_original,queriesB_original,queriesA, queriesB, hit_items_A, hit_items_B, sections_modeled, remaining_sections_A, remaining_sections_B, options):
# Initialize
verbose = options.show
output_dir = options.outdir
dummy_dir = options.dummy_dir
hydrogens = options.hbplus
force_model =options.force
python_path = config.get('Paths', 'python_path')
src_path = config.get('Paths','modppi_path')
modeller_path = os.path.join(config.get('Paths', 'modeller_path'))
modpy_path = os.path.join(src_path, config.get('Paths', 'functions_path'),"modpy")
numMod= options.nmodels
renumerate = options.renumerate
# Assign the PID to the dummy modeling and avoid overwritting files
modelling_dummy_name = 'modelling_' + str(os.getpid()) + str(random.randint(0,os.getpid()))
#modelling_dummy_name = 'modelling_' + str(os.getpid())
make_subdirs(dummy_dir, subdirs = [modelling_dummy_name])
modelling_dir = os.path.join(dummy_dir, modelling_dummy_name)
# Get items from the hits
query_A_orig = queriesA_original.get(hit_items_A[0])
query_B_orig = queriesB_original.get(hit_items_B[0])
query_A = queriesA.get(hit_items_A[0]).get_sequence()
query_B = queriesB.get(hit_items_B[0]).get_sequence()
query_name_A = hit_items_A[0]
query_name_B = hit_items_B[0]
query_id_A = query_name_A.split(':')[0]
query_start = hit_items_A[4][0]
query_end = int(hit_items_A[4][-1]) + int(hit_items_B[4][-1])
template_name_A_chain = hit_items_A[1]
template_name_B_chain = hit_items_B[1]
template_chain_A_chain = template_name_A_chain.split('_')[-1]
template_chain_B_chain = template_name_B_chain.split('_')[-1]
template_A_chain_start = hit_items_A[5][0]
template_B_chain_start = hit_items_B[5][0]
template_id_A = "_".join(template_name_A_chain.split('_')[:-1])
template_id_B = "_".join(template_name_B_chain.split('_')[:-1])
sequences_complex = {}
sequences_complex.setdefault("A",query_A_orig)
sequences_complex.setdefault("B",query_B_orig)
# Get the positions of the current section
extension_threshold = int(config.get('Parameters', 'extension_threshold'))
current_A_section = [hit_items_A[4][0], hit_items_A[4][-1]]
current_B_section = [hit_items_B[4][0], hit_items_B[4][-1]]
current_sections = [current_A_section, current_B_section]
current_interaction = '%s::%s' %(query_name_A, query_name_B)
# Initialize 'sections_modeled' dictionary
if not sections_modeled.get(current_interaction):
section_group = sections_modeled.setdefault(current_interaction, [])
section_group.append(current_sections)
# Check if the segments of the current interaction belong to a previous group
for section_pair in sections_modeled.get(current_interaction):
# The segments must be within a given interval
if (section_pair[0][0] - extension_threshold <= current_sections[0][0] <= section_pair[0][0] + extension_threshold and
section_pair[0][1] - extension_threshold <= current_sections[0][1] <= section_pair[0][1] + extension_threshold and
section_pair[1][0] - extension_threshold <= current_sections[1][0] <= section_pair[1][0] + extension_threshold and
section_pair[1][1] - extension_threshold <= current_sections[1][1] <= section_pair[1][1] + extension_threshold):
current_sections = section_pair
break
# If the segments are not within the interval, create a new group
else:
section_group = sections_modeled.setdefault(current_interaction, [])
section_group.append(current_sections)
# Get the sections that have not been used in the alignment
query_A_fragment_used = hit_items_A[2].replace('-', '')
query_B_fragment_used = hit_items_B[2].replace('-', '')
remaining_terminus_A = query_A.split(query_A_fragment_used)
remaining_terminus_B = query_B.split(query_B_fragment_used)
Nterminus_name_A = '%s_1-%s' %(query_name_A, hit_items_A[4][0] - 1)
Cterminus_name_A = '%s_%s-%s' %(query_name_A, hit_items_A[4][-1] + 1, len(query_A))
Nterminus_name_B = '%s_1-%s' %(query_name_B, hit_items_B[4][0] - 1)
Cterminus_name_B = '%s_%s-%s' %(query_name_B, hit_items_B[4][-1] + 1, len(query_B))
# If there are remaining sections, store them in the dictionary
if hit_items_A[4][0] > 1:
remaining_sections_A[Nterminus_name_A] = ProteinSequence(Nterminus_name_A, remaining_terminus_A[0])
if hit_items_A[4][-1] < len(query_A):
remaining_sections_A[Cterminus_name_A] = ProteinSequence(Cterminus_name_A, remaining_terminus_A[-1])
if hit_items_B[4][0] > 1:
remaining_sections_B[Nterminus_name_B] = ProteinSequence(Nterminus_name_B, remaining_terminus_B[0])
if hit_items_B[4][-1] < len(query_B):
remaining_sections_B[Cterminus_name_B] = ProteinSequence(Cterminus_name_B, remaining_terminus_B[-1])
#Create LOG for tests
if verbose:
dummy_log_file="%s/%s.log"%(modelling_dir, template_id_A)
dummy_log=open(dummy_log_file,"a")
# Create PDB file
if verbose: sys.stdout.write('\t\t-- Using templates %s and %s...\n' %(template_name_A_chain, template_name_B_chain))
pdb_name = template_id_A
dummy_pdb_file = '%s/%s.pdb' %(modelling_dir, pdb_name.replace(":","-"))
# Initialize PDB object
pdb_obj = PDB()
#Check template in PDB files
src_path = config.get('Paths','modppi_path')
pdb_path = os.path.join(src_path, config.get('Paths', 'pdb_path'), template_id_A[1:3].lower())
pdb_file = os.path.join(pdb_path, 'pdb' + template_id_A.lower() + '.ent')
if not os.path.exists(pdb_file):
sys.stderr.write('WARNING: PDB file %s was not found, try compressed\n' %(pdb_file))
pdb_file = os.path.join(pdb_path, 'pdb' + template_id_A.lower() + '.ent.gz')
#Check now template in 3DiD files
if not os.path.exists(pdb_file):
sys.stderr.write('WARNING: PDB file %s was not found, try 3DiD ".brk" suffix\n' %(pdb_file))
pdb_path = os.path.join(src_path, config.get('Paths', '3did_path'))
pdb_file = os.path.join(pdb_path, template_id_A.lower() + '.brk')
if not os.path.exists(pdb_file):
sys.stderr.write('WARNING: PDB file %s was not found, try 3DiD ".brk" suffix compressed\n' %(pdb_file))
pdb_file = os.path.join(pdb_path, template_id_A.lower() + '.brk.gz')
# If the PDB file is not found in the database, skips to the next interaction
if not os.path.exists(pdb_file):
sys.stderr.write('WARNING: PDB file %s was not found\n' %(pdb_file))
raise ModelException
pdb_chain_obj = PDB(pdb_file)
pdb_chain_obj.clean()
# Add only the chains present in the alignment
pdb_obj.add_chain(pdb_chain_obj.get_chain_by_id(template_chain_A_chain))
pdb_obj.add_chain(pdb_chain_obj.get_chain_by_id(template_chain_B_chain))
# Get sequences from PDB, where 'x' are gaps and 'X' are heteroatoms
pdb_seqA = pdb_obj.chains[0].gapped_protein_sequence.replace('x', '-').replace('X', '.')
pdb_seqB = pdb_obj.chains[1].gapped_protein_sequence.replace('x', '-').replace('X', '.')
# Create the dummy PDB file
pdb_obj.clean()
pdb_obj.write(output_file = dummy_pdb_file,force=True)
# Check contacts
PPI_threshold_type = config.get('Parameters', 'PPI_threshold_type')
PPI_distance_threshold = float(config.get('Parameters', 'PPI_distance_threshold'))
protein_complex = Complex(pdb_obj, PPI_type = PPI_threshold_type, PPI_distance = PPI_distance_threshold)
# If the proteins don't form a complex, avoids modelling
if len(protein_complex.PPInterfaces[0].contacts) == 0:
sys.stderr.write('WARNING: No interaction between %s and %s ( for %s %s)\n' %(template_name_A_chain, template_name_B_chain, query_name_A, query_name_B))
remove_files([dummy_pdb_file])
raise ModelException
else:
if verbose: sys.stdout.write('\t\t\t-- Accepted interaction between %s and %s ( for %s %s)...\n' %(template_name_A_chain, template_name_B_chain, query_name_A, query_name_B))
# Correct possible discrepancies between the template sequence found in the FASTA file of the nodes in the PIN and the sequence found in the PDB file
# e.g. The sequence of a protein can have an 'X' in the FASTA file and an 'M' in the newest version of the PDB file
template_seqA = hit_items_A[3]
template_seqA_ungapped = re.sub('-', '', template_seqA)
pdbA_section = pdb_seqA[hit_items_A[5][0]-1:hit_items_A[5][-1]]
for pair in itertools.izip(template_seqA_ungapped, pdbA_section):
if pair[0] == 'X' or pair[0] == 'x':
template_seqA = re.sub('[xX]', pair[1], template_seqA, 1)
template_seqB = hit_items_B[3]
template_seqB_ungapped = re.sub('-', '', template_seqB)
pdbB_section = pdb_seqB[hit_items_B[5][0]-1:hit_items_B[5][-1]]
for pair in itertools.izip(template_seqB_ungapped, pdbB_section):
if pair[0] == 'X' or pair[0] == 'x':
template_seqB = re.sub('[xX]', pair[1], template_seqB, 1)
if verbose: dummy_log.write("Hits_items_A: %s\n"%([str(x) for x in hit_items_A]))
if verbose: dummy_log.write("Hits_items_B: %s\n"%([str(x) for x in hit_items_B]))
if verbose: dummy_log.write("pdbA_section %s\n"%pdbA_section)
if verbose: dummy_log.write("pdbB_section %s\n"%pdbB_section)
if verbose: dummy_log.write("length PDB A: %d\n"%len(pdb_seqA))
if verbose: dummy_log.write("length PDB B: %d\n"%len(pdb_seqB))
# Add the remaining residues at the beginning or at the end of the template sequences, if needed
template_seqA = re.sub('[xX]', '-', template_seqA)
if template_A_chain_start > 1:
template_A_first_residues = ''.join(pdb_seqA[:hit_items_A[5][0]-1])
template_seqA = template_A_first_residues + template_seqA
if hit_items_A[5][-1] < len(pdb_seqA):
template_seqA += ''.join(pdb_seqA[hit_items_A[5][-1]:])
template_seqB = re.sub('[xX]', '-', template_seqB)
if template_B_chain_start > 1:
template_B_first_residues = ''.join(pdb_seqB[:hit_items_B[5][0]-1])
template_seqB = template_B_first_residues + template_seqB
if hit_items_B[5][-1] < len(pdb_seqB):
template_seqB += ''.join(pdb_seqB[hit_items_B[5][-1]:])
if verbose: dummy_log.write("FINAL template_seqA %s\n"%template_seqA)
if verbose: dummy_log.write("FINAL template_seqB %s\n"%template_seqB)
# Add gaps at the beginning of the query sequences, if needed
gaps_number_A_chain_beginning = 0
gaps_number_B_chain_beginning = 0
if template_A_chain_start > 1:
gaps_number_A_chain_beginning = int(template_A_chain_start) - 1
if template_B_chain_start > 1:
gaps_number_B_chain_beginning = int(template_B_chain_start) - 1
A_chain_query_seq = ''.join(['-' for i in range(gaps_number_A_chain_beginning)]) + re.sub('[xX]', '-', hit_items_A[2])
B_chain_query_seq = ''.join(['-' for i in range(gaps_number_B_chain_beginning)]) + re.sub('[xX]', '-', hit_items_B[2])
# Add gaps at the end of the query sequences, if needed
for pair in itertools.izip_longest(A_chain_query_seq, template_seqA):
if pair[0] == None:
A_chain_query_seq += '-'
for pair in itertools.izip_longest(B_chain_query_seq, template_seqB):
if pair[0] == None:
B_chain_query_seq += '-'
# Create PIR alignment
query_whole_seq = A_chain_query_seq + '/' + B_chain_query_seq + '*'
template_whole_seq = template_seqA + '/' + template_seqB + '*'
header1 = '>P1;%s\nsequence:%s:%s:.:%s:.:.:.:.:.' %(query_id_A, query_id_A, query_start, query_end)
header2 = '>P1;%s\nstructureX:%s:1:%s:.:%s:.:.:.:.' %(template_id_A.replace(":","-"), template_id_A.replace(":","-"), template_chain_A_chain, template_chain_B_chain)
lines = []
lines.append(header1)
lines.extend([query_whole_seq[i:i+60] for i in range(0, len(query_whole_seq), 60)])
lines.append(header2)
lines.extend([template_whole_seq[i:i+60] for i in range(0, len(template_whole_seq), 60)])
pir_alignment = '\n'.join(lines)
pir_file = open('%s/alignment.pir' %(modelling_dir), 'w+')
for line in lines:
pir_file.write('%s\n' %(line))
pir_file.close()
# Model
# Create a folder for the models of each type of interaction
if '-' in query_name_A:
query_name_A = query_name_A.rsplit('_', 1)[0]
if '-' in query_name_B:
query_name_B = query_name_B.rsplit('_', 1)[0]
interaction_dir = os.path.join(output_dir , '%s::%s' %(query_name_A, query_name_B))
if not os.path.exists(interaction_dir):
make_subdirs(output_dir, subdirs = ['./%s::%s' %(query_name_A, query_name_B)])
# If the models do not yet exist, proceed and add in the list of MODELS
do_model=False
model_path = os.path.abspath(interaction_dir)
for imodel in xrange(1,numMod+1):
model_name = '%s_%s_%d-%d::%s_%s_%d-%d#%d.pdb' %(template_id_A, template_chain_A_chain, current_sections[0][0], current_sections[0][1], template_id_B, template_chain_B_chain, current_sections[1][0], current_sections[1][1],imodel)
model_path_model = os.path.join(model_path , model_name)
#print "CHECK %s %s\n"%(do_model,model_path_model)
with open(interaction_dir + '/%s.list' %(current_interaction), 'a+') as paths_to_models_file:
if model_path_model not in paths_to_models_file.read(): paths_to_models_file.write(model_path_model + '\n')
if not do_model and not fileExist( model_path_model ): do_model=True
#Complete the set of models
if do_model or force_model:
# Keep the current working directory, then change to the modelling folder
cwd = os.getcwd()
os.chdir(modelling_dir)
try:
if options.optimize:
process = subprocess.check_output([os.path.join(modeller_path, 'modpy.sh'), os.path.join(python_path, 'python'), os.path.join(modpy_path, 'simpleModel.py'), '--pir=' + './alignment.pir', '--out=%s-%s' %(template_name_A_chain, template_name_B_chain), '--models=%d'%(numMod), '--optimize'], stderr = subprocess.STDOUT)
else:
process = subprocess.check_output([os.path.join(modeller_path, 'modpy.sh'), os.path.join(python_path, 'python'), os.path.join(modpy_path, 'simpleModel.py'), '--pir=' + './alignment.pir', '--out=%s-%s' %(template_name_A_chain, template_name_B_chain), '--models=%d'%(numMod)], stderr = subprocess.STDOUT)
except Exception as e:
sys.stderr.write("ERROR: %s\n"%(e))
sys.stderr.write("LOCATION; %s\n"%modelling_dir)
if verbose: os.system("grep get_ran %s"%(template_name_A_chain+"-"+template_name_B_chain+".log"))
if verbose: sys.stderr.write("\t\tSkip models with template %s\n"%(model_name))
os.chdir(cwd)
raise ModelException
# Clean and rename all models
for imodel in xrange(1,numMod+1):
label_model=99990000+imodel
input_model = '%s.B%s.pdb' %(query_id_A,str(label_model))
model_name = '%s_%s_%d-%d::%s_%s_%d-%d#%d.pdb' %(template_id_A, template_chain_A_chain, current_sections[0][0], current_sections[0][1], template_id_B, template_chain_B_chain, current_sections[1][0], current_sections[1][1],imodel)
model_path_model = os.path.join(model_path , model_name)
if fileExist(os.path.abspath('%s' %(input_model))):
# Check contacts
check_pdb_obj=PDB(os.path.abspath('%s' %(input_model)))
PPI_threshold_type = config.get('Parameters', 'PPI_threshold_type')
PPI_distance_threshold = float(config.get('Parameters', 'PPI_distance_threshold'))
check_protein_complex = Complex(check_pdb_obj, PPI_type = PPI_threshold_type, PPI_distance = PPI_distance_threshold)
if len(check_protein_complex.PPInterfaces[0].contacts) == 0:
if verbose: sys.stdout.write("\t\t\t-- Skip model without contacts %s\n"%model_name)
continue
else:
if verbose: sys.stdout.write("\t\t\t-- Accepted model %s\n"%model_name)
if hydrogens:
if verbose: sys.stdout.write("\t\t\t-- Adding hydrogens and relaxing the model %s\n"%model_name)
output_model=model_name
try:
add_hydrogens(config,os.path.abspath("./"),input_model, output_model,dummy_dir)
except ValueError as e:
sys.stderr.write("WARNING %s\n"%e)
os.rename(input_model, output_model)
else:
output_model=model_name
os.rename(input_model, output_model)
if renumerate:
if verbose: sys.stdout.write("\t\t\t-- Renumerate residues as original sequence\n")
output_model_renumber=model_name+".re"
try:
pdb_renumber=PDB()
pdb_renumber=renumber_pdb(config,os.path.abspath("./"),output_model,sequences_complex,os.path.abspath("./"))
pdb_renumber.write(output_model_renumber)
os.rename(output_model_renumber,output_model)
except Exception as e:
sys.stderr.write("WARNING %s\n"%e)
shutil.copy(output_model, model_path_model)
os.chdir(cwd)
try:
shutil.rmtree(modelling_dir)
except Exception as e:
sys.stderr.write("WARNING first attempt to remove folder %s\n"%e)
try:
os.system("\\rm -r %s"%(modelling_dir))
except Exception as ee:
sys.stderr.write("WARNING last attempt %s\n"%ee)
return sections_modeled, remaining_sections_A, remaining_sections_B