def generate_aotoencoder_traindata():
file = open('pdbname.txt')
errorlog = open('wrong_pdb.txt', 'a')
writer1 = tf.python_io.TFRecordWriter(
'autoencoder_traindata/train.tfrecords')
writer2 = tf.python_io.TFRecordWriter(
'autoencoder_traindata/test.tfrecords')
lines = file.readlines()
count = 0
for line in lines:
count += 1
print count, line.strip()
arg = ['pdbfile=samermax_pisa_pdb/' + line.strip()]
try:
cube = np.zeros((32, 32, 32))
cube[:31, :31, :31] = pdb2voxel.run(arg)
cube = cube.astype(int)
cube_list = cube.flatten()
example = tf.train.Example(features=tf.train.Features(
feature={
"data":
tf.train.Feature(int64_list=tf.train.Int64List(
value=cube_list)), #require a list of int
}))
if count % 5 == 0:
writer2.write(example.SerializeToString())
else:
writer1.write(example.SerializeToString())
except:
errorlog.write(line)
writer1.close()
writer2.close()
errorlog.close()
file.close()
def create_tfrecords():
file=open('pdbname.txt')
errorlog=open('wrong_pdb.txt','a')
writer=tf.python_io.TFRecordWriter('train.tfrecords')
writer2=tf.python_io.TFRecordWriter('test.tfrecords')
lines=file.readlines()
count=0
for line in lines:
count+=1
print count,line.split('\n')[0]
arg=['pdbfile='+line.split('\n')[0]]
try:
cube=pdb2voxel.run(arg)
cube=cube.astype(int)
cube_list=cube.flatten()
example = tf.train.Example(features=tf.train.Features(feature={
"data": tf.train.Feature(int64_list=tf.train.Int64List(value=cube_list)), #require a list of int
}))
if count%5==0:
writer2.write(example.SerializeToString())
else:
writer1.write(example.SerializeToString())
except:
errorlog.write(line)
writer1.close()
writer2.close()
errorlog.close()
file.close()
def generate_gene():
f = open('pdbname.txt')
pdbnames = f.readlines()
f.close()
pdbnames = np.array(pdbnames)
pdbnames = np.sort(pdbnames)
saved_model_path = 'train_autoencoder/model'
in_tensor_find, z_tensor_find, out_tensor_find = auto_encoder_z.generate_session(
1)
saver = tf.train.Saver()
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
model_path = tf.train.latest_checkpoint(saved_model_path)
saver.restore(sess, model_path)
for pdbname in pdbnames:
try:
pdbname = pdbname.strip()
inpdbfile = 'samermax_pisa_pdb/' + pdbname
outpdbfile = 'pisa_data_gene/' + pdbname.split('.')[0] + '.txt'
voxel = pdb2voxel.run(['pdbfile=%s' % inpdbfile])
in_ = np.zeros(shape=(1, 32, 32, 32, 1))
in_[0, :31, :31, :31, 0] = voxel
z_ = sess.run(z_tensor_find[0],
feed_dict={in_tensor_find[0]: in_})
z_ = z_.reshape((-1, 1))
np.savetxt('%s' % outpdbfile, z_, fmt='%.3f')
except:
print '%s error!!!!!!!!!!!!!!!!!!!!!!!!!!!!' % pdbname
def generate_samermax_pdb(pdbname):
#pisa
try:
pdbname = pdbname.strip()
#cavitymodel=model_interface.build_model(pdbfile,'pdb',20,None)
voxel = pdb2voxel.run(['pdbfile=rotate_pisa_pdb/%s' % pdbname])
voxel2pdb.write_pdb(voxel, 'samermax_pisa_pdb/%s' % pdbname, 70)
except:
print '%s error!!!!!!!!!!!!!!!!!!!!!!!!!!!!' % pdbname
fw = open('wrong_samples.txt', 'a')
fw.write('%s\n' % pdbname)
fw.close()
def write2pdb(group, rmax, output_folder):
num = len(group)
os.system('mkdir %s/sub2' % output_folder)
os.system('mkdir %s/sub3' % output_folder)
for ii in range(num):
voxel2pdb.write_pdb(group[ii], '%s/sub2/%d.pdb' % (output_folder, ii),
rmax)
fix = '%s/sub2/0.pdb' % output_folder
data = []
for ii in range(num):
mov = '%s/sub2/%d.pdb' % (output_folder, ii)
align.run(fix, mov, '%s/sub3/%d.pdb' % (output_folder, ii))
voxel = pdb2voxel.run(['pdbfile=%s/sub3/%d.pdb' % (output_folder, ii)])
data.append(voxel)
data = np.array(data)
data = np.mean(data, axis=0)
data = np.greater(data, 0.5).astype(int)
voxel2pdb.write_pdb(data, '%s/out.pdb' % output_folder, rmax)
os.system('rm -rf %s/sub2' % output_folder)
os.system('rm -rf %s/sub3' % output_folder)
def generate_iq(filename):
#pisa
iq_file = 'saxs_sample.txt'
iq_data = np.loadtxt(iq_file, delimiter=' ', dtype=float)
filename = filename.strip()
try:
t_voxel = pdb2voxel.run(['pdbfile=samermax_pisa_pdb/%s' % filename])
#cavitymodel=model_interface.build_model(inpdbfile,'pdb',20,None)
#rmax=cavitymodel.rmax*0.9
rmax = 70
iq_t_curve, t_exp_data = map2iq.run_get_voxel_iq(
t_voxel, iq_file, rmax)
iq_t_curve = np.array(iq_t_curve)
iq_t_curve = iq_t_curve / iq_t_curve[0]
cacl_iq = np.concatenate(
[iq_data[:, 0].reshape(-1, 1),
iq_t_curve.reshape(-1, 1)], axis=1)
np.savetxt('pisa_data_iq/%s.iq' % filename.split(".")[0],
cacl_iq,
fmt='%f')
except:
print "%serror!!!!!!!!!!!!!!!!" % filename
def write2pdb(group, rmax, output_folder, iq_file=None, target_pdb=None):
if iq_file is not None:
tar_iq_curve = np.loadtxt(iq_file, usecols=(0))
tar_iq_curve = tar_iq_curve.reshape(-1, 1)
num = len(group)
os.system('mkdir %s/sub2' % output_folder)
os.system('mkdir %s/sub3' % output_folder)
for ii in range(num):
voxel2pdb.write_pdb(group[ii], '%s/sub2/%d.pdb' % (output_folder, ii),
rmax)
fix = '%s/sub2/0.pdb' % output_folder
data = []
for ii in range(num):
mov = '%s/sub2/%d.pdb' % (output_folder, ii)
align.run(fix, mov, '%s/sub3/%d.pdb' % (output_folder, ii))
voxel = pdb2voxel.run(['pdbfile=%s/sub3/%d.pdb' % (output_folder, ii)])
data.append(voxel)
data = np.array(data)
data = np.mean(data, axis=0)
ccp4data = np.copy(data)
data = np.greater(data, 0.3).astype(int)
if iq_file is not None:
iq_curve, exp_data = map2iq.run_get_voxel_iq(data, iq_file, rmax)
iq_curve = np.array(iq_curve)
#iq_curve=iq_curve/iq_curve[0]
newiq_curve = np.concatenate((tar_iq_curve, iq_curve.reshape((-1, 1))),
axis=1)
np.savetxt('%s/final_saxs.txt' % output_folder, newiq_curve)
voxel2pdb.write_pdb(data, '%s/out.pdb' % output_folder, rmax)
'''
if iq_file is not None:
out_voxel=pdb2voxel.run(['pdbfile=%s/out.pdb'%output_folder])
out_curve,exp_data=map2iq.run_get_voxel_iq(out_voxel,iq_file,rmax)
out_curve=np.array(out_curve)
out_curve=out_curve/out_curve[0]
newout_curve=np.concatenate((tar_iq_curve,out_curve.reshape((-1,1))),axis=1)
np.savetxt('%s/final_pdb_saxs.txt'%output_folder,newout_curve)
'''
ccp4data = flex.double(ccp4data)
pdb2zernike.ccp4_map_type(ccp4data,
15,
rmax / 0.9,
file_name='%s/out.ccp4' % output_folder)
shiftrmax = rmax / 0.9
args = [
'fix=%s/out.ccp4' % output_folder, 'typef=ccp4',
'mov=%s/out.pdb' % output_folder,
'rmax=%f' % shiftrmax
]
zalign.run(args, output_folder)
os.system('rm -r %s/sub2' % output_folder)
os.system('rm -r %s/sub3' % output_folder)
if target_pdb is not None:
args = [
'fix=%s/out.ccp4' % output_folder, 'typef=ccp4',
'mov=%s' % target_pdb,
'rmax=%f' % shiftrmax
]
zalign.run(args, output_folder)
if 'sample.pdb' in os.listdir(output_folder):
args = [
'fix=%s/out.ccp4' % output_folder, 'typef=ccp4',
'mov=%s/sample.pdb' % output_folder,
'rmax=%f' % shiftrmax
]
zalign.run(args, output_folder)