tissue_names = []
num_internal_nodes = 0
X = []
y = []
enhancers_list = list(enhancers_dict.keys())
for name in nodes:
if name == 'internal node':
num_internal_nodes += 1
else:
tissue_names.append(name)
print(tissue_names)
parent_path_mat = build_parent_path_mat(pc_mat)
num_edges = len(parent_path_mat)
delta_mat = np.zeros(shape=(embedding_size, num_edges))
root_vector = np.zeros(shape=embedding_size)
tissue_dfs = {}
for t in tissue_names:
t_df = pd.read_csv(os.path.join('data_files', 'CT_enhancer_features_matrices',
t + '_enhancer_features_matrix.csv'), index_col='cCRE_id')
t_df = t_df.loc[enhancers_list]
tissue_dfs[t] = t_df
for enhancer in enhancers_list:
X.append(get_one_hot_encoding(enhancers_dict[enhancer]))
for row in samples_df.itertuples():
added_in_X_and_y = False
genome_id = getattr(row, 'ID')
for i, examples_list in enumerate(node_examples):
if genome_id in examples_list:
if not added_in_X_and_y:
phenotype = eval(getattr(row, 'Phenotype'))[0] # the y value
features = eval(getattr(row, 'Features')) # the x value
y.append(phenotype)
X.append(features)
added_in_X_and_y = True
mapping.append((example_number, i))
if added_in_X_and_y:
example_number += 1
parent_path_tensor = build_parent_path_mat(parent_child)
num_features = len(X[0])
num_nodes = len(parent_child[0])
num_edges = len(parent_path_tensor)
root_weights = np.zeros(shape=num_features)
edge_tensor_matrix = np.zeros(shape=(num_features, num_edges))
test_auc_output = []
val_auc_output = []
average_time_seed = 0
for s in args.seeds:
init_time = time.time()
print('New seed: ' + str(s))
if node_list[-1].left is not None:
node_queue.append(node_list[-1].left)
if node_list[-1].right is not None:
node_queue.append(node_list[-1].right)
num_nodes = len(node_list)
num_edges = num_nodes - 1
# constructing the parent-child matrix, would be nice to find a faster way to do this
parent_child_mat = np.zeros(shape=(num_nodes, num_nodes),
dtype=np.float32)
for child_idx in range(1,
len(node_list)): # excluding the root
parent_idx = node_list.index(node_list[child_idx].parent)
parent_child_mat[parent_idx, child_idx] = 1.0
pp_mat = build_parent_path_mat(parent_child_mat,
num_edges=num_edges)
# split the leaves into train and test
train_idx, valid_idx = split_indices(range(len(leaves)))
# constructing train and valid x and y matrices
train_col_idx = [leaves[i][0] for i in train_idx]
valid_col_idx = [leaves[i][0] for i in valid_idx]
train_col_idx_tensor = torch.tensor(train_col_idx,
device=device)
valid_col_idx_tensor = torch.tensor(valid_col_idx,
device=device)
train_x = torch.tensor(np.asarray(
[leaves[i][1].x for i in train_idx]),
device=device,