def process_test_data(protein_path = 'test_protein',ligand_path = 'test_ligand',o_path = 'test_output'):
for ligand_file in os.listdir(ligand_path):
ligand_st_path = os.path.join(ligand_path,ligand_file)#Ligand file path
protein_st_path = os.path.join(protein_path,ligand_file)#protein file path
print ligand_st_path,protein_st_path
if not os.path.exists(protein_st_path):#protein file not exist
#print protein_st_path,'not exist'
continue
else:
ligand_st = structure.StructureReader(ligand_st_path).next()
protein_st = structure.StructureReader(protein_st_path).next()
complex_id=ligand_file[:4]
try:#dangerous
fp_gen_path = fp_gen1.gen_fp(receptor=protein_st,binder=ligand_st,complex_id=complex_id,root_path=o_path)
except:
print complex_id,'cannot be processed'
sift_path = os.path.join(complex_id,'pattern.dat')
if os.path.exists(sift_path):
print complex_id,'is processed'
continue
avg_sift_path = os.path.join(o_path,sift_path)
avg_sift.gen_avg_sift(fp_gen_path,avg_sift_path)#generate average sift
def get_avg_matrix(receptor_path='',ligands_path = [],output_fp_path = '',output_pattern_path = ''):
#get receptor
receptor = structure.StructureReader(receptor_path).next()
receptor.title = receptor_path
print receptor_path,'set'
rec_tree = fp_gen1.distance_tree()
rec_tree.set_receptor_structure(receptor)
rec_tree.parse_receptor()
#get ligands
ligands = []
for ligand_path in ligands_path:
lig = structure.StructureReader(ligand_path).next()
lig.title = ligand_path
ligands.append(lig)
print ligand_path,'added'
for lig in ligands:
rec_tree.find_close_residues(lig)
print lig,'close residues found'
with open(output_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 'finger print',output_fp_path,'saved'
avg_sift.gen_avg_sift(output_fp_path,output_pattern_path)
print 'pattern',output_pattern_path,'saved'
def cal_avg_sift_from_complex(complex_path='data/HL_chain/1RD8-1918/complex.1000.pdb'):
print complex_path
_,chain_name,complex_id,instance_name = complex_path.split('/')
p_chain_path=os.path.join('processed_data',chain_name)
if not os.path.exists(p_chain_path):
os.mkdir(p_chain_path)
p_complex_path=os.path.join(p_chain_path,complex_id)
if not os.path.exists(p_complex_path):
os.mkdir(p_complex_path)
p_instance_path=os.path.join(p_complex_path,instance_name)[:-4]
if not os.path.exists(p_instance_path):
os.mkdir(p_instance_path)
antigen_path = os.path.join(p_instance_path,'antigen.pdb')
antibody_path = os.path.join(p_instance_path,'antibody.pdb')
print antigen_path,antibody_path
antibody,antigen = split_protein_protein_complex_manual(complex_path = complex_path,antigen_path = antigen_path ,antibody_path=antibody_path)
fp_path = os.path.join(p_instance_path,'fp.out')
print antibody,antigen
try:
fp_gen_path = fp_gen1.gen_fp(receptor=antibody,binder = antigen,fp_path= fp_path)#get the finger print
except:
print 'failed\n'
return
else:
print 'good\n'
sift_path=os.path.join(p_instance_path,'avg_sift.out')
print sift_path
avg_sift.gen_avg_sift(fp_gen_path,sift_path)#generate average sift
def cal_avg_sift_from_complex(complex_path='data/HL_chain/1RD8-1918/complex.1000.pdb',output_dir="/home/xiaohan/Desktop",split_fun=globals()['split_protein_protein_complex_manual']):
if not os.path.exists(output_dir):
os.mkdir(output_dir);
binder_path = os.path.join(output_dir,'binder.pdb')
receptor_path = os.path.join(output_dir,'receptor.pdb')
split_fun(complex_path = complex_path,binder_path = binder_path ,receptor_path=receptor_path,chain_info_file = os.path.join(os.path.dirname(complex_path),u"新建文本文档.txt"))
fp_path = os.path.join(output_dir,'fp.out')
fp_gen_path = fp_gen1.gen_fp(receptor_file=receptor_path,binder_file = binder_path,fp_path= fp_path)#get the finger print
#fp_gen_path = fp_gen1.gen_fp(receptor_file = binder_path,binder_file = receptor_path,fp_path= fp_path)#get the finger print
sift_path=os.path.join(output_dir,'avg_sift.out')
avg_sift.gen_avg_sift(fp_gen_path,sift_path)#generate average sift
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
if self.used_bits_number != 13:
out_sift = ""
for chunk in self.sifts[lig_name].sift:
for bit in self.active_bits:
out_sift.append(chunk.bit_set[bit])
return "".join(out_sift)
else:
"""
concatenate the bit string
"""
out_sift = []
for fp_chunk in self.sifts[lig_name].sift.keys():
out_sift.append("".join(self.sifts[lig_name].sift[fp_chunk].bit_set))
# for bit in self.sifts[lig_name].sift[fp_chunk].bit_set:
# out_sift += str(bit)
return "".join(out_sift)
if __name__ == "__main__":
rec_file = "/home/xiaohan/Downloads/protein/1CE1/1CE1_antibody.pdb"
bind_file = "/home/xiaohan/Downloads/protein/1CE1/1CE1_antigen.pdb"
from avg_sift import gen_avg_sift
cutoff = 4.0
fp_file = "/home/xiaohan/Desktop/1CE1_fp_%d.dat" % (cutoff)
pat_file = "/home/xiaohan/Desktop/1CE1_pat_%d.dat" % (cutoff)
gen_fp(rec_file, bind_file, fp_file, cutoff=cutoff)
gen_avg_sift(fp_file, pat_file)
fp_fp = "%s/fp" %data_root_fp
avg_sift_fp = "%s/avg_sift" %data_root_fp
ligand_fp = "%s/ligand" %data_root_fp
binder_fp= "%s/binder" %data_root_fp
#create them
prepare_dirs([fp_fp , avg_sift_fp , ligand_fp , binder_fp])
#calculation start
for complex_fp in glob.glob(os.path.join(pdb_fp,'*')):
name_with_pdb = os.path.split(complex_fp)[-1]
complex_id = os.path.split(complex_fp)[-1].split('.')[0]
#init paths
cur_binder_fp = os.path.join(binder_fp , name_with_pdb)
cur_ligand_fp = os.path.join(ligand_fp , name_with_pdb)
cur_avg_sift_fp = os.path.join(avg_sift_fp , "%s.sift" %complex_id)
cur_fp_fp = os.path.join(fp_fp , "%s.fp" %complex_id)
#split complex into ligand and binder
split_complex(complex_fp , cur_binder_fp , cur_ligand_fp , has_atom_or_hetatm, ATOM_as_binder)
antibody , antigen = load_structure(cur_binder_fp) , load_structure(cur_ligand_fp)
#generate finger print
fp_gen_path = gen_fp(receptor=antibody,binder = antigen,fp_path= cur_fp_fp)#get the finger print
#generate sift
gen_avg_sift(cur_fp_fp,cur_avg_sift_fp)#generate average sift