def plot_vertex_weight_histogram(max_distance):
"""
This function plots one histogram for the vertex weights and another for the chromosomes with the breaks
:param max_distance:
:return:
"""
from matplotlib import pyplot as plt
# Iterate over the files
all_weights = np.array([], dtype=np.float)
all_labels = np.array([])
for file_name in os.listdir(data_path):
# Load the brakes
breaks, list_of_pairs = load_breaks(os.path.join(data_path, file_name))
if len(breaks) == 0:
print 'WARNING: Empty data file', file_name
shutil.move(os.path.join(data_path, file_name), os.path.join('../../data/raw_original_data/empty_files', file_name))
continue
# Generate the vertices and edges
adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights = generateGraph(breaks, list_of_pairs,
max_distance)
all_weights = np.append(all_weights, np.array(vertex_weights))
all_labels = np.append(all_labels, np.array(vertex_labels))
from collections import Counter
all_w_dict = Counter(all_weights.astype(np.float))
little = {k: all_w_dict[k] for k in all_w_dict.keys() if all_w_dict[k] >= 100}
print(all_w_dict)
all_lab_dict = sorted(Counter(all_labels).items(), key=lambda (k, v): (natural_key(k), v))
x, y = zip(*all_lab_dict)
# print np.array(all_w_dict.values()).astype(np.float)/np.max(all_w_dict.values())
plt.figure()
plt.bar(list(all_w_dict.keys()), np.array(all_w_dict.values()).astype(np.float) / np.max(all_w_dict.values()),
color='b')
plt.title('Weights ' + str(max_distance))
plt.savefig('../../data/plots/test_weights' + str(max_distance) + '.png')
plt.figure()
plt.bar(list(little.keys()), np.array(little.values()).astype(np.float) / np.max(little.values()),
color='b')
plt.title('Weights little ' + str(max_distance))
plt.savefig('../../data/plots/test_weights_little' + str(max_distance) + '.png')
plt.figure()
plt.bar(x, y, color='g')
plt.title('Labels ' + str(max_distance))
plt.savefig('../../data/plots/test_labels' + str(max_distance) + '.png')
# plt.show()
plt.close()
def plot_one_file(file_name, max_distance=1000):
# Load the brakes
breaks, list_of_pairs = load_breaks(os.path.join(data_path, file_name))
if len(breaks) == 0:
print 'WARNING: Empty data file', file_name
# Generate the vertices and edges
adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights = generateGraph(breaks, list_of_pairs, max_distance)
# Create the graph
g = generateNXGraph(adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights, self_links=False,
connected_only=True)
# Print the graph
print 'Showing graph of ', file_name
printGraph(g, name=file_name, visualize=False, show_vertex_weights=False)
def plot_all(max_distance):
# Iterate over the files
for file_name in os.listdir(data_path):
# Load the brakes
breaks, list_of_pairs = load_breaks(os.path.join(data_path, file_name))
if len(breaks) == 0:
print 'WARNING: Empty data file', file_name
continue
# Generate the vertices and edges
adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights = generateGraph(breaks, list_of_pairs,
max_distance)
# Create the graph
g = generateNXGraph(adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights, self_links=False,
connected_only=True)
# Print the graph
print 'Showing graph of ', file_name
printGraph(g, show_vertex_weights=False)
print vertex_ranges
def generate_all_patient_graphs(max_distance, data_path, output_path):
subgraphs = []
# Iterate over the files
count = 0
for filename in os.listdir(data_path):
breaks, list_of_pairs = load_breaks(os.path.join(data_path, filename))
if len(breaks) == 0:
# print 'WARNING: Empty data file',file_name
continue
adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights = generateGraph(
breaks, list_of_pairs, max_distance)
g = generateNXGraph(adjacency_matrix,
vertex_labels,
vertex_ranges,
vertex_weights,
self_links=False,
connected_only=True)
candidates = list(nx.connected_component_subgraphs(g))
for c in candidates:
if len(c.nodes()) >= 3 and len(c.nodes()) >= 3:
print count, filename
subgraphs.append(c)
count += 1
# Iterate over subgraphs and store
with open(output_path + '/gspan_subgraphs_w' + str(max_distance) + '.txt',
'w') as f:
counter = 0
for g in subgraphs:
f.write('t # ' + str(counter) + '\n')
counter += 1
# Iterate over vertices
for v in g.nodes():
f.write('v ' + str(v) + ' 2\n')
# Iterate over all edges
for edge in g.edges():
f.write('e ' + str(edge[0]) + ' ' + str(edge[1]) + ' 2\n')
def generate_all_patient_graphs(max_distance):
"""
Generates graphs according to the window sizes for all files.
Decompose these into subgraphs, according to connected components, and stores in a format compatible for gSpan.
"""
subgraphs = []
# Iterate over the files
count = 0
for filename in os.listdir(PATIENTS_PATH):
breaks, list_of_pairs = load_breaks(os.path.join(PATIENTS_PATH, filename))
if len(breaks) == 0:
# print 'WARNING: Empty data file',file_name
continue
adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights = generateGraph(breaks, list_of_pairs,
max_distance)
g = generateNXGraph(adjacency_matrix, vertex_labels, vertex_ranges, vertex_weights, self_links=False,
connected_only=True)
candidates = list(nx.connected_component_subgraphs(g))
for c in candidates:
if len(c.nodes()) >= 3 and len(c.nodes()) >= 3:
print count, filename
subgraphs.append(c)
count += 1
# Iterate over subgraphs and store
with open(GSPAN_COMPATIBLE_GRAPHS_PATH + 'gspan_subgraphs_w' + str(max_distance) + '.txt', 'w') as f:
counter = 0
for g in subgraphs:
f.write('t # ' + str(counter) + '\n')
counter += 1
# Iterate over vertices
for v in g.nodes():
f.write('v ' + str(v) + ' 2\n')
# Iterate over all edges
for edge in g.edges():
f.write('e ' + str(edge[0]) + ' ' + str(edge[1]) + ' 2\n')
#counter_empty = 0
cliques_3 = []
cliques_4 = []
data_path = '../vcfshorts/allfiles'
#data_path = '../vcfshorts/chromexsamples'
#file_name = 'fca3f7d0-2231-661c-e040-11ac0c4832fd.vcf.tsv'
#file_name = 'feccee20-a62d-4152-b832-b9fdaca87a61.vcf_chromex.tsv'
for file_name in os.listdir(data_path):
if file_name.split('.')[0] in cancer_types[t]:
breaks, list_of_pairs = load_breaks(os.path.join(data_path,file_name))
#if len(breaks) == 0:
# #print 'WARNING: Empty data file',file_name
# counter_empty +=1
adjacency_matrix, vertex_labels, vertex_ranges = generateGraph(breaks, list_of_pairs, max_distance=100)
counter += 1
if len(vertex_labels)!=0:
c3,c4 = getCliques(adjacency_matrix,vertex_labels)
cliques_3.append(c3)
cliques_4.append(c4)
else:
cliques_3.append(0)
cliques_4.append(0)
#print 'Cancer type',t, 'Mean/std of 3 cliques',np.asarray(cliques_3).mean(),np.asarray(cliques_3).std(),'Mean/std of 4 cliques',np.asarray(cliques_4).mean(),np.asarray(cliques_4).std()
print t, counter, np.around(np.asarray(cliques_3).mean(),decimals=2),np.around(np.asarray(cliques_3).std(),decimals=2),np.around(np.asarray(cliques_4).mean(),decimals=2),np.around(np.asarray(cliques_4).std(),decimals=2)
counter_all += counter
#counter_all_empty += counter_empty
cliques_all_3 += cliques_3
cliques_all_4 += cliques_4