def main():
with lt.open_file() as write_f:
for f in lt.files_in_dir(sys.argv[-1]):
pdb_id = os.path.split(f)[1].split('.')[0]
ligands = check_ligands(f)
if len(ligands) > 0:
print >> write_f, pdb_id, '\t', ' '.join(ligands)
def main():
hbs = []
for hb2_f in lt.files_in_dir(sys.argv[-1]):
if hb2_f[-4:] == '.hb2':
hbs.extend(read_hb2(hb2_f))
write_initial(hbs)
hbs = [(pro, phos_res, neighbor_res)
for pro, phos_res, neighbor_res, _ in hbs]
lt.pickle_dump(hbs, 'hbplus')
def main():
null = open(os.devnull, 'w')
# sys.stderr = null
"""
"""
res_neighbors = []
for pdb_f in lt.files_in_dir(sys.argv[-1]):
f_path, f_name = os.path.split(pdb_f)
f_name, f_exten = os.path.splitext(f_name)
if f_exten == '.pdb':
pdb_id = f_name
res_neighbors.extend(pdb_neighbors(pdb_f, pdb_id))
write_result(res_neighbors, file_suffix='original')
sta = [len(v) for k, v in res_neighbors.iteritems()]
sta = lt.lis_sta(sta)
with lt.open_file('phos_neighbors_num') as w_f:
for k, v in sta.items():
print >> '{0:<10}{1:<10}'.format(k, v)
def main():
null = open(os.devnull, 'w')
sys.stderr = null
dist = []
for pdb_f in lt.files_in_dir(sys.argv[-1]):
pdb_id = pdb_f[-8:-4]
p_dist = pdb_dist(pdb_f, pdb_id)
dist += p_dist
def myround(n):
n = np.round(n, 1)
a, b = str(n).split('.')
n = a + '.' + b[0]
return float(n)
dist = sorted(dist)
dist = [myround(d) for d in dist]
dist = [d for d in dist if d < 10.0]
sta = lt.lis_sta(dist)
with lt.open_file(file_suffix='dist_sta') as w_f:
for num, freq in sta:
print >> w_f, '{0:<10}{1}'.format(num, freq)
def read_pickle():
for f in lt.files_in_dir(sys.argv[-1]):
if 'hbplus' in f:
hbplus_neighbors = lt.pickle_load(f)
if 'salt' in f:
salt_neighbors = lt.pickle_load(f)
hbplus_neighbors_dic = OrderedDict()
for pdb_id, res, neighbors in hbplus_neighbors:
hbplus_neighbors_dic[(pdb_id, res)] = neighbors
salt_neighbors_dic = {}
for pdb_id, res, neighbors in salt_neighbors:
salt_neighbors_dic[(pdb_id, res)] = neighbors
combine_neighbors = []
for k, v in hbplus_neighbors_dic.iteritems():
if k in salt_neighbors_dic.keys():
v.extend(salt_neighbors_dic[k])
v = set(v)
combine_neighbors.append((k[0], k[1], v))
else:
combine_neighbors.append((k[0], k[1], v))
return combine_neighbors
def main():
null = open(os.devnull, 'w')
# sys.stderr = null
"""
write original center:neighbors
write pdb_ids
write neighbors:pdbs
"""
def write_result(res_neighbors, file_suffix):
pdbs = list(set([pdb_id for pdb_id, _, _ in res_neighbors]))
with lt.open_file(file_name='pdb_id',
inner_dir=file_suffix,
file_suffix=file_suffix) as w_f:
print >> w_f, 'num:', '\t', len(pdbs)
for pdb in pdbs:
print >> w_f, pdb
res_neighbors = sorted(res_neighbors)
with lt.open_file(file_name='res_neighbors',
inner_dir=file_suffix,
file_suffix=file_suffix) as w_f:
for pdb_id, res, neighbors in res_neighbors:
print >> w_f, '{0:<20}{1:<20}{2}'.format(
pdb_id, res, ' '.join(neighbors))
HASH = {
'VAL': 'V',
'ILE': 'I',
'LEU': 'L',
'GLU': 'E',
'GLN': 'Q',
'ASP': 'D',
'ASN': 'N',
'HIS': 'H',
'TRP': 'W',
'PHE': 'F',
'TYR': 'Y',
'ARG': 'R',
'LYS': 'K',
'SER': 'S',
'THR': 'T',
'MET': 'M',
'ALA': 'A',
'GLY': 'G',
'PRO': 'P',
'CYS': 'C'
}
neighbor_pdbs = {}
for pdb_id, res, neighbors in res_neighbors:
neighbors = [n.split('_')[0] for n in neighbors]
neighbors = [HASH[n] for n in neighbors]
neighbors = ''.join(sorted(neighbors))
if not neighbors in neighbor_pdbs.keys():
neighbor_pdbs[neighbors] = [pdb_id]
else:
if not pdb_id in neighbor_pdbs[neighbors]:
neighbor_pdbs[neighbors].append(pdb_id)
neighbor_pdbs = [(len(pdbs), neighbors, pdbs)
for neighbors, pdbs in neighbor_pdbs.iteritems()]
neighbor_pdbs = sorted(neighbor_pdbs, reverse=True)
with lt.open_file(file_name='neighbor_pdbs',
inner_dir=file_suffix,
file_suffix=file_suffix) as w_f:
for _, neighbors, pdbs in neighbor_pdbs:
pdbs = [pdb for pdb in pdbs]
print >> w_f, '{0:<10}{1}'.format(neighbors, ' '.join(pdbs))
res_neighbors = []
for pdb_f in lt.files_in_dir(sys.argv[-1]):
f_path, f_name = os.path.split(pdb_f)
f_name, f_exten = os.path.splitext(f_name)
pdb_id = f_name
res_neighbors.extend(pdb_neighbors(pdb_f, pdb_id))
#write original
pdbs = set([
pdb_id for pdb_id, _, neighbors in res_neighbors if len(neighbors) > 0
])
with lt.open_file(file_suffix='original_pdbs') as w_f:
for p in pdbs:
print >> w_f, p
with lt.open_file(file_suffix='orignal') as w_f:
for pdb_id, res, neighbors in res_neighbors:
if len(neighbors) > 0:
print >> w_f, '{0:<20}{1:<20}{2}'.format(
pdb_id, res, ' '.join(neighbors))
with lt.open_file(file_suffix='orignal_simple') as w_f:
for pdb_id, res, neighbors in res_neighbors:
if len(neighbors) > 0:
HASH = {
'VAL': 'V',
'ILE': 'I',
'LEU': 'L',
'GLU': 'E',
'GLN': 'Q',
'ASP': 'D',
'ASN': 'N',
'HIS': 'H',
'TRP': 'W',
'PHE': 'F',
'TYR': 'Y',
'ARG': 'R',
'LYS': 'K',
'SER': 'S',
'THR': 'T',
'MET': 'M',
'ALA': 'A',
'GLY': 'G',
'PRO': 'P',
'CYS': 'C'
}
neighbors = [n.split('_')[0] for n in neighbors]
neighbors = [HASH.get(n, '') for n in neighbors]
neighbors = ''.join(sorted(neighbors))
print >> w_f, '{0:<20}{1:<20}{2}'.format(
pdb_id, res, ''.join(neighbors))
# res_neighbors format: [(pdb_id,residue,[neighbor_resiudes])]
# delete non-standard residues
STAND_RES = [
'VAL', 'ILE', 'LEU', 'GLU', 'GLN', 'ASP', 'ASN', 'HIS', 'TRP', 'PHE',
'TYR', 'ARG', 'LYS', 'SER', 'THR', 'MET', 'ALA', 'GLY', 'PRO', 'CYS'
]
res_neighbors = [(pdb_id, res,
[n for n in neighbors if n.split('_')[0] in STAND_RES])
for pdb_id, res, neighbors in res_neighbors]
res_neighbors = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) > 0]
# bad_pdbs = set([pdb_id for pdb_id,res,neighbors in res_neighbors for n in neighbors if not n.split('_')[0] in STAND_RES])
# res_neighbors = [(pdb_id,res,neighbors) for pdb_id,res,neighbors in res_neighbors if not pdb_id in bad_pdbs]
write_result(res_neighbors, file_suffix='filter_non_standard')
# # delete water
# res_neighbors = [(pdb_id, res, [n for n in neighbors if n.split(
# '_')[-1] != 'HOH']) for pdb_id, res, neighbors in res_neighbors]
# res_neighbors = [(pdb_id, res, neighbors) for pdb_id,
# res, neighbors in res_neighbors if len(neighbors) > 0]
# write_result(res_neighbors, file_suffix='delete_water')
# delete same_chain
res_neighbors = [
(pdb_id, res,
[n for n in neighbors if n.split('_')[-1] != res.split('_')[-1]])
for pdb_id, res, neighbors in res_neighbors
]
res_neighbors = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) > 0]
write_result(res_neighbors, file_suffix='filter_same_chain')
#delete same_chain
# bad_pdbs = set([pdb_id for pdb_id,res,neighbors in res_neighbors for n in neighbors if n.split('_')[-1] == res.split('_')[-1]])
# res_neighbors = [(pdb_id,res,neighbors) for pdb_id,res,neighbors in res_neighbors if not pdb_id in bad_pdbs]
# write_result(res_neighbors, file_suffix='delete_same_chain')
#neighbors = 1
res_neighbors1 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 1]
write_result(res_neighbors1, file_suffix='1_neighbor')
#neighbors = 2
res_neighbors2 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 2]
write_result(res_neighbors2, file_suffix='2_neighbor')
#neighbors = 3
res_neighbors3 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 3]
write_result(res_neighbors3, file_suffix='3_neighbor')
#neighbors = 4
res_neighbors4 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 4]
write_result(res_neighbors4, file_suffix='4_neighbor')
#neighbors = 5
res_neighbors5 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 5]
write_result(res_neighbors5, file_suffix='5_neighbor')
#neighbors = 6
res_neighbors6 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 6]
write_result(res_neighbors6, file_suffix='6_neighbor')
#neighbors > 6
res_neighbors7 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) > 6]
write_result(res_neighbors7, file_suffix='7_neighbor')
num1 = len(set([pdb_id for pdb_id, _, _ in res_neighbors1]))
num2 = len(set([pdb_id for pdb_id, _, _ in res_neighbors2]))
num3 = len(set([pdb_id for pdb_id, _, _ in res_neighbors3]))
num4 = len(set([pdb_id for pdb_id, _, _ in res_neighbors4]))
num5 = len(set([pdb_id for pdb_id, _, _ in res_neighbors5]))
num6 = len(set([pdb_id for pdb_id, _, _ in res_neighbors6]))
num7 = len(set([pdb_id for pdb_id, _, _ in res_neighbors7]))
sta = [num1, num2, num3, num4, num5, num6, num7]
with lt.open_file(file_suffix='sta') as w_f:
print >> w_f, '{0:<20}{1}'.format('neighbors num', 'pdb_num')
for i, num in enumerate(sta):
print >> w_f, '{0:<20}{1}'.format(i + 1, num)
print >> w_f, '{0:<20}{1}'.format('total', sum(sta))
def main():
phos_neighbors = {}
good_pdb = []
bad_pdb = []
for pdb_f in lt.files_in_dir(sys.argv[-1]):
f_path, f_name = os.path.split(pdb_f)
f_id, f_ex = os.path.splitext(f_name)
neighbors = pdb_neighbors(pdb_f, f_id)
print neighbors
if len(neighbors) > 0:
for k, v in neighbors.iteritems():
if len(v) > 0:
phos_neighbors[f_id] = neighbors
good_pdb.append(f_id)
else:
bad_pdb.append(f_id)
lt.write_list(good_pdb, 'good_pdb')
lt.write_list(bad_pdb, 'bad_pdb')
# delete res_id and clear empty
phos_neighbors_v1 = {}
for k, v in phos_neighbors.iteritems():
phos_neighbors_v1[k] = {}
for vk, vv in v.iteritems():
phos_neighbors_v1[k][vk] = []
vvn = []
if len(vv) > 0:
for vi in vv:
vvn.append(vi.split('_')[0])
vvn = sorted(vvn)
phos_neighbors_v1[k][vk] = vvn
# delete non-standard residues
stand_res = [
'VAL', 'ILE', 'LEU', 'GLU', 'GLN', 'ASP', 'ASN', 'HIS', 'TRP', 'PHE',
'TYR', 'ARG', 'LYS', 'SER', 'THR', 'MET', 'ALA', 'GLY', 'PRO', 'CYS'
]
phos_neighbors_v2 = {}
for k, v in phos_neighbors_v1.iteritems():
phos_neighbors_v2[k] = {}
for vk, vv in v.iteritems():
vvn = []
for vi in vv:
if vi in stand_res:
vvn.append(vi)
if len(vvn) > 0:
vvn = sorted(vvn)
phos_neighbors_v2[k][vk] = vvn
# delete uncomplete patterns
phos_neighbors_v3 = {}
for k, v in phos_neighbors_v2.iteritems():
phos_neighbors_v3[k] = {}
for vk, vv in v.iteritems():
if len(vv) >= 3:
phos_neighbors_v3[k][vk] = vv
def write_phosneighbors(phos_neighbors, ofile):
keys = phos_neighbors.keys()
keys = sorted(keys)
for k in keys:
print >> ofile, k
print >> ofile, '*' * 80
for k in keys:
v = phos_neighbors[k]
print >> ofile, k
for vk, vv in v.iteritems():
print >> ofile, ' ', vk, '\t', ' '.join(vv)
print >> ofile, '*' * 80
phos_sta = {}
for k, v in phos_neighbors.iteritems():
for vk, vv in v.iteritems():
if not tuple(vv) in phos_sta.keys():
phos_sta[tuple(vv)] = [k]
else:
phos_sta[tuple(vv)].append(k)
phos_sta_list = []
for k, v in phos_sta.iteritems():
phos_sta_list.append((len(v), k, v))
phos_sta_list = sorted(phos_sta_list, reverse=True)
for len_v, k, v in phos_sta_list:
print >> ofile, '{0:150}{1:<8}{2}'.format(k, len_v, ','.join(v))
ofile = lt.open_file('phos_neighbors_original')
write_phosneighbors(phos_neighbors, ofile)
ofile = lt.open_file('phos_neighbors_sorted')
write_phosneighbors(phos_neighbors_v1, ofile)
ofile = lt.open_file('phos_neighbors_clear')
write_phosneighbors(phos_neighbors_v2, ofile)
ofile = lt.open_file('phos_neighbors_clean')
write_phosneighbors(phos_neighbors_v3, ofile)
import os
import sys
import subprocess
import lt
for f in lt.files_in_dir(sys.argv[-1]):
subprocess.call('~/hbplus/hbplus/hbplus', f, "-f 'SEP.txt'")
def main():
null = open(os.devnull, 'w')
# sys.stderr = null
"""
write original center:neighbors
write pdb_ids
write neighbors:pdbs
"""
res_neighbors = []
for pdb_f in lt.files_in_dir(sys.argv[-1]):
f_path, f_name = os.path.split(pdb_f)
f_name, f_exten = os.path.splitext(f_name)
if f_exten == '.pdb':
pdb_id = f_name
res_neighbors.extend(pdb_neighbors(pdb_f, pdb_id))
write_result(res_neighbors, file_suffix='original')
#neighbors = 1
res_neighbors1 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 1]
write_result(res_neighbors1, file_suffix='1_neighbor')
#neighbors = 2
res_neighbors2 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 2]
write_result(res_neighbors2, file_suffix='2_neighbor')
#neighbors = 3
res_neighbors3 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 3]
write_result(res_neighbors3, file_suffix='3_neighbor')
#neighbors = 4
res_neighbors4 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 4]
write_result(res_neighbors4, file_suffix='4_neighbor')
#neighbors = 5
res_neighbors5 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 5]
write_result(res_neighbors5, file_suffix='5_neighbor')
#neighbors = 6
res_neighbors6 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) == 6]
write_result(res_neighbors6, file_suffix='6_neighbor')
#neighbors > 6
res_neighbors7 = [(pdb_id, res, neighbors)
for pdb_id, res, neighbors in res_neighbors
if len(neighbors) > 6]
write_result(res_neighbors7, file_suffix='7_neighbor')
num1 = len(set([pdb_id for pdb_id, _, _ in res_neighbors1]))
num2 = len(set([pdb_id for pdb_id, _, _ in res_neighbors2]))
num3 = len(set([pdb_id for pdb_id, _, _ in res_neighbors3]))
num4 = len(set([pdb_id for pdb_id, _, _ in res_neighbors4]))
num5 = len(set([pdb_id for pdb_id, _, _ in res_neighbors5]))
num6 = len(set([pdb_id for pdb_id, _, _ in res_neighbors6]))
num7 = len(set([pdb_id for pdb_id, _, _ in res_neighbors7]))
sta = [num1, num2, num3, num4, num5, num6, num7]
with lt.open_file(file_suffix='sta') as w_f:
print >> w_f, '{0:<20}{1}'.format('neighbors num', 'pdb_num')
for i, num in enumerate(sta):
print >> w_f, '{0:<20}{1}'.format(i + 1, num)
print >> w_f, '{0:<20}{1}'.format('total', sum(sta))