def del_temp_K_prot(DB_version, DB_type, delete):
"""
Check (and -optional- delete) LAkernel output files which are not completed.
Parameters
----------
DB_version : str
string of the DrugBank version number
format : "drugbank_vX.X.X" exemple : "drugbank_v5.1.1"
DB_type : str
string of the DrugBank type exemple: "S0h"
delete : boolean
whether or not to delete the file
Returns
-------
None
"""
# data_dir variable
data_dir = 'data/' + DB_version + '/' + DB_type + '/'
# kernels directory
kernels_dir = root + data_dir + 'kernels/'
# get the DBdataBase preprocessed
preprocessed_DB = get_DB(DB_version, DB_type)
dict_ind2prot = preprocessed_DB.proteins.dict_ind2prot
nb_prot = preprocessed_DB.proteins.nb
list_ = []
for index in range(nb_prot):
# output_filename
dbid = dict_ind2prot[index]
output_filename = kernels_dir + 'LAkernel/LA_' + DB_type + \
'_' + dbid + '.txt'
if os.path.isfile(output_filename):
output_file = open(output_filename)
count_nb_line = 0
for line in output_file.readlines():
count_nb_line += 1
if count_nb_line != nb_prot - index:
list_.append(output_filename)
else:
list_.append(output_filename)
print(list_)
print('delete', delete)
if delete == True:
for outf in list_:
if os.path.isfile(outf):
os.remove(outf)
def check_temp_K_prot(DB_version, DB_type):
"""
Check and process make_temp_K_prot() for the proteins for which the \
LAkernel has not been processed.
See the description of make_temp_K_prot for more details
Parameters
----------
DB_version : str
string of the DrugBank version number
format : "drugbank_vX.X.X" exemple : "drugbank_v5.1.1"
DB_type : str
string of the DrugBank type exemple: "S0h"
Returns
-------
None
"""
# data_dir variable
data_dir = 'data/' + DB_version + '/' + DB_type + '/'
# kernels directory
kernels_dir = root + data_dir + 'kernels/'
# get the DBdataBase preprocessed
preprocessed_DB = get_DB(DB_version, DB_type)
dict_ind2prot = preprocessed_DB.proteins.dict_ind2prot
nb_prot = preprocessed_DB.proteins.nb
list_ = []
for index in range(nb_prot):
# output_filename
dbid = dict_ind2prot[index]
output_filename = kernels_dir + 'LAkernel/LA_' + DB_type + \
'_' + dbid + '.txt'
if not os.path.isfile(output_filename):
list_.append(dbid)
print("list of uncompleted proteins", list_)
# data_dir variable
data_dir = 'data/' + args.DB_version + '/' + args.DB_type + '/'
if not os.path.exists(root + data_dir + '/' + 'cross_validation/kronSVM'):
os.mkdir(root + data_dir + '/' + 'cross_validation/kronSVM')
print("kronSVM cross validation directory for", args.DB_type, ",",
args.DB_version, "created.")
else:
print("kronSVM cross validation directory for", args.DB_type, ",",
args.DB_version, "already exists.")
cv_dirname = root + data_dir + 'cross_validation/'
kronsvm_cv_dirname = root + data_dir + 'cross_validation/kronSVM/'
preprocessed_DB = get_DB(args.DB_version, args.DB_type)
kernels = get_K_mol_K_prot(args.DB_version, args.DB_type, args.center_norm,
args.norm)
# # Get the train datasets
# train_folds = get_train_folds(args.DB_version, args.DB_type)
# nb_folds = len(train_folds)
# nb_clf = len(train_folds[0])
# Get the nested folds
nested_cv_dirname = root + data_dir + 'cross_validation/nested_folds/'
nested_folds_array_filename = nested_cv_dirname + args.DB_type + '_nested_folds_array.data'
nested_folds_array = pickle.load(open(nested_folds_array_filename, 'rb'))
# data_dir variable
data_dir = 'data/' + args.DB_version + '/' + args.DB_type + '/'
if not os.path.exists(root + data_dir + '/' + 'cross_validation/NRLMF'):
os.mkdir(root + data_dir + '/' + 'cross_validation/NRLMF')
print("NRLMF cross validation directory for", args.DB_type, ",", args.DB_version,
"created.")
else:
print("NRLMF cross validation directory for", args.DB_type, ",", args.DB_version,
"already exists.")
cv_dirname = root + data_dir + 'cross_validation/'
nrlmf_cv_dirname = root + data_dir + 'cross_validation/NRLMF/'
DB = get_DB(args.DB_version, args.DB_type)
kernels = get_K_mol_K_prot(args.DB_version, args.DB_type, center_norm=True, norm=False)
DB_drugs_kernel = kernels[0]
DB_proteins_kernel = kernels[1]
# Get the nested folds
nested_cv_dirname = root + data_dir + 'cross_validation/nested_folds/'
if args.balanced_on_proteins == True:
if args.balanced_on_drugs == True:
nested_folds_array_filename = nested_cv_dirname \
+ args.DB_type + '_nested_folds_double_balanced_1_clf_array.data'
output_filename = nrlmf_cv_dirname + args.DB_type + \
'_NRLMF_cv_nested_double_balanced_1_clf_pred.data'
else:
def make_group_K_prot(DB_version, DB_type):
"""
Process the similarity between all the proteins with LAkernel
Use make_K_mol.center_and_normalise_kernel()
Write 2 files:
- ..._K_prot.data : LA Kernel
- ..._K_prot_norm.data : Centered and Normalised Kernel
Parameters
----------
DB_version : str
string of the DrugBank version number
format : "drugbank_vX.X.X" exemple : "drugbank_v5.1.1"
DB_type : str
string of the DrugBank type
Returns
-------
None
"""
# data_dir variable
data_dir = 'data/' + DB_version + '/' + DB_type + '/'
# kernels directory
kernels_dir = root + data_dir + 'kernels/'
# get the DBdataBase preprocessed
preprocessed_DB = get_DB(DB_version, DB_type)
dict_ind2prot = preprocessed_DB.proteins.dict_ind2prot
nb_prot = preprocessed_DB.proteins.nb
X = np.zeros((nb_prot, nb_prot))
for i in range(nb_prot):
# output_filename
dbid = dict_ind2prot[i]
output_filename = kernels_dir + 'LAkernel/LA_' + DB_type + \
'_' + dbid + '.txt'
j = i
for line in open(output_filename, 'r'):
r = float(line.rstrip())
X[i, j] = r
X[j, i] = X[i, j]
j += 1
if j != nb_prot:
print(dbid, 'kernel, corresponding to the protein nb ', i,
', is uncompleted')
# normalized or unnormalized
for norm_type in ['norm', 'unnorm']:
if norm_type == 'unnorm':
kernel_filename = kernels_dir + DB_type + '_K_prot.data'
pickle.dump(X, open(kernel_filename, 'wb'), protocol=2)
elif norm_type == 'norm':
K_norm = center_and_normalise_kernel(X)
kernel_filename = kernels_dir + DB_type + '_K_prot_norm.data'
pickle.dump(K_norm, open(kernel_filename, 'wb'), protocol=2)
print(X[100, 100], K_norm[100, 100])
print(X[100, :], K_norm[100, :])
print(X[500, 500], K_norm[500, 500])
print(X[500, :], K_norm[500, :])
def make_temp_K_prot(DB_version, DB_type, index):
"""
Process the similarity of one particular protein with the others
Process the similarity (thanks to the L(ocal)A(lignment) Kernel) of \
the protein (with the key *index* in the dict_ind2prot dictionary, and \
corresponding to the fasta FASTA_A) with the proteins between *index+1*\
and *nb_prot* (corresponding to fasta FASTA_B) in the dict_ind2prot \
dictionary, with the command: \
'LAkernel_direct FASTA_A FASTA B'
Then append the output to the file LA_kernel/LA_..._[dbid].txt, where dbid \
is the value of the key *index* in the dict_ind2prot dictionary.
Parameters
----------
DB_version : str
string of the DrugBank version number
format : "drugbank_vX.X.X" exemple : "drugbank_v5.1.1"
DB_type : str
string of the DrugBank type exemple: "S0h"
index : int
index of the protein in the dictionaries
Returns
-------
None
"""
# data_dir variable
data_dir = 'data/' + DB_version + '/' + DB_type + '/'
# kernels directory
kernels_dir = root + data_dir + 'kernels/'
#create LAkernel directory
if not os.path.exists(kernels_dir + 'LAkernel/'):
os.mkdir(kernels_dir + 'LAkernel/')
print("LAkernel directory for", data_dir, "created")
# get the DataBase preprocessed
preprocessed_DB = get_DB(DB_version, DB_type)
dict_protein = preprocessed_DB.proteins.dict_protein
dict_ind2prot = preprocessed_DB.proteins.dict_ind2prot
# output_filename
dbid = dict_ind2prot[index]
output_filename = kernels_dir + 'LAkernel/LA_' + DB_type + \
'_' + dbid + '.txt'
nb_prot = preprocessed_DB.proteins.nb
FASTA1 = dict_protein[dbid]
if not os.path.isfile(output_filename):
print(index, ":", dbid)
for j in range(index, nb_prot):
dbid2 = dict_ind2prot[j]
FASTA2 = dict_protein[dbid2]
com = LAkernel_path + ' ' + FASTA1 + ' ' + FASTA2 + \
' >> ' + output_filename
cmd = os.popen(com)
cmd.read()
print("completed")
from DTI_prediction.process_dataset.process_DB import get_DB from DTI_prediction.process_dataset.DB_utils import Drugs, Proteins, Couples, FormattedDB from DTI_prediction.process_dataset.DB_utils import check_drug, check_protein, check_couple, get_couples_from_array root = '../CFTR_PROJECT/' DB_version = "drugbank_v5.1.5" DB_type = "S0h" process_name = "VMO" # pattern_name variable pattern_name = DB_type + '_' + process_name # data_dir variable data_dir = 'data/' + DB_version + '/' + DB_type + '/' + pattern_name + '/' preprocessed_DB = get_DB(DB_version, DB_type) from DTI_prediction.process_dataset.correct_interactions import get_orphan ivacaftor_dbid = 'DB08820' corrected_DB = copy.deepcopy(preprocessed_DB) corrected_DB = get_orphan(DB=corrected_DB, dbid=ivacaftor_dbid) # Get the kernels from DTI_prediction.make_kernels.get_kernels import get_K_mol_K_prot from DTI_prediction.make_kernels.make_K_mol import center_and_normalise_kernel kernels = get_K_mol_K_prot(DB_version, DB_type, norm=False) K_mol = kernels[0] K_prot = kernels[1]
