def _frame_loader(yaml_file, prt, key_list, indices, save_trj, fname=None):
traj_list=[]
#if not an iterator make it
if type(indices)!=list and indices.shape==(2,):
indices = [indices]
for ind in indices:
traj_index, frame_index = ind
traj_name = key_list[traj_index]
traj_list.append(load_frame(yaml_file["base_dir"],
prt.name, yaml_file["protein_dir"],
traj_name, frame_index))
trj = traj_list[0] + traj_list[1:]
if save_trj and fname is not None:
with enter_protein_mdl_dir(yaml_file, prt.name):
trj.save_xtc("%s"%fname)
if not os.path.isfile("prot.pdb"):
trj[0].save_pdb("prot.pdb")
return trj
def _frame_loader(yaml_file, prt, key_list, indices, save_trj, fname=None):
traj_list = []
#if not an iterator make it
if type(indices) != list and indices.shape == (2, ):
indices = [indices]
for ind in indices:
traj_index, frame_index = ind
traj_name = key_list[traj_index]
traj_list.append(
load_frame(yaml_file["base_dir"], prt.name,
yaml_file["protein_dir"], traj_name, frame_index))
trj = traj_list[0] + traj_list[1:]
if save_trj and fname is not None:
with enter_protein_mdl_dir(yaml_file, prt.name):
trj.save_xtc("%s" % fname)
if not os.path.isfile("prot.pdb"):
trj[0].save_pdb("prot.pdb")
return trj
def _get_common_features(yaml_file, featurizer, aligned_dict,save_df=True):
"""
Function to get the common features across protein using the common residues.
can optionally save the pandas data to the mdl_dir
:param yaml_file: The protein yaml_file
:param featurizer: featurizer object used.
:param prt_mapping: Mapping of each residue to its sequence
:param aligned_dict : Dictionary of alignments for each protein
:return:
"""
result_dict = {}
df_dict={}
for protein in yaml_file["protein_list"]:
print(protein)
#reset the featurizer
featurizer = clone(featurizer)
trj = load_random_traj(yaml_file, protein)
df = pd.DataFrame(featurizer.describe_features(trj))
prt_mapping, prt_seq = _map_residue_ind_seq_ind(yaml_file, protein,
aligned_dict[protein])
feature_vec =[]
#for every feature
for i in df.iterrows():
#get the index and the feature itself
feature_ind, feature_dict = i
all_res_in_algn = []
mapped_index_list=[]
for aa_ind in feature_dict["resids"]:
aa_code = prt_seq[aa_ind]
#make sure we have the same residue
assert(trj.top.residue(aa_ind).code==aa_code)
#get the mapping for that aa to the main alignment
mapped_index = prt_mapping[aa_ind]
#for every protein in the alignment, check if we have the same residue
#at the same position
all_res_in_algn.append(np.alltrue([aligned_dict[prt][mapped_index]==aa_code
for prt in yaml_file["protein_list"]]))
mapped_index_list.append(mapped_index)
#to account for additions and deletions, we check if the difference between
#the mapping and the actual residue codes is the same.
mapped_index_difference = [x - mapped_index_list[i - 1]
for i, x in enumerate(mapped_index_list) if i > 0]
resid_index_difference = [x - feature_dict["resids"][i - 1]
for i, x in enumerate(feature_dict["resids"]) if i > 0]
if not np.all(mapped_index_difference==resid_index_difference):
all_res_in_algn.append(False)
if np.alltrue(all_res_in_algn):
feature_vec.append(feature_ind)
df_dict[protein] = df.iloc[feature_vec]
result_dict[protein] = feature_vec
if save_df:
new_df = df.iloc[feature_vec]
with enter_protein_mdl_dir(yaml_file, protein):
verbosedump(new_df, os.path.join("feature_descriptor.h5"))
with enter_protein_data_dir(yaml_file, protein):
verbosedump(new_df, os.path.join("sliced_feature_dir",
"feature_descriptor.h5"))
return result_dict, df_dict
def _get_common_features(yaml_file, featurizer, aligned_dict,save_df=True):
"""
Function to get the common features across protein using the common residues.
can optionally save the pandas data to the mdl_dir
:param yaml_file: The protein yaml_file
:param featurizer: featurizer object used.
:param prt_mapping: Mapping of each residue to its sequence
:param aligned_dict : Dictionary of alignments for each protein
:return:
"""
result_dict = {}
df_dict={}
for protein in yaml_file["protein_list"]:
print(protein)
#reset the featurizer
featurizer = clone(featurizer)
trj = load_random_traj(yaml_file, protein)
df = pd.DataFrame(featurizer.describe_features(trj))
prt_mapping, prt_seq = _map_residue_ind_seq_ind(yaml_file, protein,
aligned_dict[protein])
feature_vec =[]
#for every feature
for i in df.iterrows():
#get the index and the feature itself
feature_ind, feature_dict = i
all_res_in_algn = []
mapped_index_list=[]
for aa_ind in feature_dict["resids"]:
aa_code = prt_seq[aa_ind]
#make sure we have the same residue
assert(trj.top.residue(aa_ind).code==aa_code)
#get the mapping for that aa to the main alignment
mapped_index = prt_mapping[aa_ind]
#for every protein in the alignment, check if we have the same residue
#at the same position
all_res_in_algn.append(np.alltrue([aligned_dict[prt][mapped_index]==aa_code
for prt in yaml_file["protein_list"]]))
mapped_index_list.append(mapped_index)
#to account for additions and deletions, we check if the difference between
#the mapping and the actual residue codes is the same.
mapped_index_difference = [x - mapped_index_list[i - 1]
for i, x in enumerate(mapped_index_list) if i > 0]
resid_index_difference = [x - feature_dict["resids"][i - 1]
for i, x in enumerate(feature_dict["resids"]) if i > 0]
if not np.all(mapped_index_difference==resid_index_difference):
all_res_in_algn.append(False)
if np.alltrue(all_res_in_algn):
feature_vec.append(feature_ind)
df_dict[protein] = df.iloc[feature_vec]
result_dict[protein] = feature_vec
if save_df:
new_df = df.iloc[feature_vec]
with enter_protein_mdl_dir(yaml_file, protein):
verbosedump(new_df, os.path.join("feature_descriptor.h5"))
with enter_protein_data_dir(yaml_file, protein):
verbosedump(new_df, os.path.join("sliced_feature_dir",
"feature_descriptor.h5"))
return result_dict, df_dict
