def split_protein_molecule_complex_manual(st_path,root_dir = 'processed_data_protein_molecule'):
"""
@params:complex structure file path
@return receptor,small molecule
"""
complex_id,_ = fileutils.splitext(os.path.basename(st_path))
root_dir
if not os.path.exists(root_dir):
os.mkdir(root_dir)
result_save_path = os.path.join(root_dir,complex_id)
if not os.path.exists(result_save_path):
os.mkdir(result_save_path)
rec_fname = '%s/%s_receptor.pdb' %(result_save_path,complex_id)
bind_fname = '%s/%s_binder.pdb' %(result_save_path,complex_id)
with open(rec_fname,'w') as rec:
with open( bind_fname,'w') as bind:
with open(st_path) as f:
for line in f.readlines():
if start_with_atom(line):
rec.write(line)
elif belongs_to_molecule(line):
bind.write(line)
rec_id,_ = fileutils.splitext(os.path.basename(rec_fname))
bind_id,_ = fileutils.splitext(os.path.basename(bind_fname))
rec = structure.StructureReader(rec_fname).next()
rec.title = rec_id
bind = structure.StructureReader(bind_fname).next()
bind.title = bind_id
return rec,bind
def split_protein_protein_complex_manual(st_path,root_dir):
"""
@params:complex structure file path
@return antibody structure, antigen structure
"""
complex_id,_ = fileutils.splitext(os.path.basename(st_path))
if not os.path.exists(root_dir):
os.mkdir(root_dir)
result_save_path = os.path.join(root_dir,complex_id)
if not os.path.exists(result_save_path):
os.mkdir(result_save_path)
ab_fname = '%s/%s_antibody.pdb' %(result_save_path,complex_id)
ag_fname = '%s/%s_antigen.pdb' %(result_save_path,complex_id)
with open(ab_fname,'w') as ab_f:
with open( ag_fname,'w') as ag_f:
with open(st_path) as f:
for line in f.readlines():
if start_with_atom(line):
if belongs_to_antibody(line):
ab_f.write(line)
else:
ag_f.write(line)
antibody_id,_ = fileutils.splitext(os.path.basename(ab_fname))
antigen_id,_ = fileutils.splitext(os.path.basename(ag_fname))
antibody = structure.StructureReader(ab_fname).next()
antibody.title = antibody_id
antigen = structure.StructureReader(ag_fname).next()
antigen.title = antigen_id
return antibody,antigen
def split_protein_protein_complex_manual(complex_path='',antigen_path='',antibody_path=''):
"""
@params:complex structure file path,to be saved antigen path,to be saved antibody path
@return antibody structure, antigen structure
"""
with open(antibody_path,'w') as ab_f:
with open(antigen_path,'w') as ag_f:
with open(complex_path) as f:
for line in f.readlines():
if start_with_atom(line):
if belongs_to_antibody(line):
ab_f.write(line[:60]+'\n')
else:
ag_f.write(line[:60]+'\n')
antibody_id,_ = fileutils.splitext(os.path.basename(antibody_path))
antigen_id,_ = fileutils.splitext(os.path.basename(antigen_path))
antibody = structure.StructureReader(antibody_path).next()
antibody.title = antibody_id
antigen = structure.StructureReader(antigen_path).next()
antigen.title = antigen_id
return antibody,antigen
def gen_protein_protein_complex_avg_sift(complex_st_path,processed_data_path = 'processed_data'):
complex_id, ext = fileutils.splitext(os.path.basename(complex_st_path))
sift_path = '%s/%s/%s_pattern.dat' %(processed_data_path,complex_id,complex_id)
if os.path.exists(sift_path):
print complex_id,'is processed'
return
antibody, antigen = split_protein_protein_complex_manual(complex_st_path,processed_data_path)#load complex structure and split it
fp_gen_path = fp_gen1.gen_fp(receptor=antibody,binder = antigen,complex_id = complex_id,root_path = processed_data_path)#get the finger print
avg_sift.gen_avg_sift(fp_gen_path,sift_path)#generate average sift
with open(fp_path, 'w') as out_fp:
for key in rec_tree.fingerprints.sifts.keys():
rec_tree.fingerprints.fill_missing_zeros(rec_tree.min_res, rec_tree.max_res, key)
fp_string = rec_tree.fingerprints.get_sift_string(key)
out_fp.write(rec_tree.receptor.title + ':' + key + ':' + str(rec_tree.min_res) + ':' + fp_string + '\n')
print 'saved to',fp_path
return fp_path
if __name__ == "__main__":
rec_file = 'protein2protein/1A2Y_antigen.pdb'
bind_file = 'protein2protein/1A2Y_antibody.pdb'
basename, ext = fileutils.splitext(os.path.basename(rec_file))
rec_tree = distance_tree()
#receptor data
rec = structure.StructureReader(rec_file).next()
if rec.title == "":
rec.title = basename
rec_tree.set_receptor_structure(rec)
rec_tree.parse_receptor()
#bind data
basename, ext = fileutils.splitext(os.path.basename(rec_file))
bind = structure.StructureReader(bind_file).next()
