def train_ppi_matrix_classifier(input_files, result_file):
with open(ppi_train_interactions_file, "rb") as f:
labels = cPickle.load(f)
X, Y = ppi_learning.predictorI_aggregated_data([ppi_matrices_train_file],
labels)
np.random.seed(42)
ppi_learning.cv_experiment(X, Y, labels, result_file)
def test_existing_ppi_matrix_classifier(input_files, result_file):
with open(ppi_train_interactions_file, "rb") as f:
labels = cPickle.load(f)
X, Y = ppi_learning.predictorI_aggregated_data([ppi_matrices_train_file],
labels)
np.random.seed(42)
clf = joblib.load(ppi_classifier_file)
Y_score = clf.predict_proba(X)[:, 1]
Y_pred = np.array(Y_score > 0.5, dtype=int)
accuracy = [metrics.accuracy_score(Y, Y_pred)]
precision = [metrics.precision_score(Y, Y_pred)]
recall = [metrics.recall_score(Y, Y_pred)]
auc = [metrics.roc_auc_score(Y, Y_score)]
TP = sum((Y_pred == 1)[Y == 1])
FP = sum((Y_pred == 1)[Y == 0])
TN = sum((Y_pred == 0)[Y == 0])
FN = sum((Y_pred == 0)[Y == 1])
accuracy = np.array(accuracy)
precision = np.array(precision)
recall = np.array(recall)
auc = np.array(auc)
with open(result_file, "w") as f:
f.write("& " + " & ".join([
"{0:.2f}".format(a.mean(), a.std())
for a in [accuracy, precision, recall, auc]
]) + "\\\\\n")
f.write("TP={0} FP={1} TN={2} FN={3}".format(TP, FP, TN, FN))
def train_ppi_matrix_classifier(input_files, result_file):
with open(ppi_train_interactions_file, "rb") as f:
labels = cPickle.load(f)
X, Y = ppi_learning.predictorI_aggregated_data([ppi_matrices_train_file],
labels)
np.random.seed(42)
ppi_learning.cv_experiment(X, Y, labels, result_file)