def run(args):
# Load unpermuted network.
edge_list = load_edge_list(args.edge_list_file, unweighted=True)
# Permute network.
G = nx.Graph()
G.add_edges_from(edge_list)
if args.seed is not None:
random.seed(args.seed)
minimum_swaps = int(math.ceil(args.Q*G.number_of_edges()))
if not args.connected:
G = nx.double_edge_swap(G, minimum_swaps, 2**30)
else:
# If G is not connected, then we perform the connected double edge swap algorithm on a
# largest connected component of G.
if not nx.is_connected(G):
G = max(nx.connected_component_subgraphs(G), key=len)
# The current connected double edge swap algorithm does not guarantee a minimum number of
# successful edge swaps, so we enforce it.
current_swaps = 0
while current_swaps<minimum_swaps:
remaining_swaps = max(minimum_swaps-current_swaps, 100)
additional_swaps = nx.connected_double_edge_swap(G, remaining_swaps)
current_swaps += additional_swaps
permuted_edge_list = G.edges()
# Save permuted_network.
save_edge_list(args.permuted_edge_list_file, permuted_edge_list)
def run(args):
# Load data.
index_to_gene, gene_to_index = load_index_gene(args.index_gene_file)
edge_list = load_edge_list(args.edge_list_file,
index_to_gene,
unweighted=True)
gene_to_score = load_gene_score(args.gene_score_file)
# Find degrees of network genes in subgraph induced by scored genes.
G = nx.Graph()
G.add_edges_from(edge_list)
G = G.subgraph(gene_to_score)
largest_cc = max(nx.connected_components(G), key=len)
G = G.subgraph(largest_cc)
common_genes = set(G.nodes)
degree_to_nodes = defaultdict(set)
for node in common_genes:
degree = G.degree(node)
degree_to_nodes[degree].add(node)
distinct_degrees = set(degree_to_nodes)
# Bin genes by degree.
bins = list()
current_bin = list()
for degree in sorted(distinct_degrees, reverse=True):
current_bin += sorted(degree_to_nodes[degree])
if len(current_bin) >= args.min_size:
bins.append(current_bin)
current_bin = list()
if bins:
bins[-1] += current_bin
elif current_bin:
bins.append(current_bin)
# Save degree bins.
with open(args.output_file, 'w') as f:
f.write('\n'.join('\t'.join(current_bin) for current_bin in bins))
def run(args):
# Load edge list.
edge_list = load_edge_list(args.edge_list_file)
# Construct adjacency matrix.
k = min(min(edge[:2]) for edge in edge_list)
l = max(max(edge[:2]) for edge in edge_list)
A = np.zeros((l - k + 1, l - k + 1), dtype=np.float64)
if args.directed:
for i, j, weight in edge_list:
A[j - k, i - k] = weight
else:
for i, j, weight in edge_list:
A[i - k, j - k] = A[j - k, i - k] = weight
# Choose beta.
if args.beta is None:
beta = balanced_beta(A, args.threshold, args.num_digits)
elif 0 < args.beta < 1:
beta = args.beta
else:
raise ValueError('{} invalid; beta must satisfy 0 < beta < 1.'.format(
args.beta))
# Construct Hierarchical HotNet similarity matrix.
P = hh_similarity_matrix(A, beta)
# Save results.
if args.output_file is not None:
save_matrix(args.output_file, P, args.name)
if args.beta_output_file is not None:
fmt = '{:.' + str(args.num_digits) + 'f}'
with open(args.beta_output_file, 'w') as f:
f.write(fmt.format(beta))
def cut_hierarchy(edge_list_file, index_gene_file, cut_height):
T = load_edge_list(edge_list_file)
index_to_gene, gene_to_index = load_index_gene(index_gene_file)
clusters = find_clusters(T, index_to_gene, cut_height)
return clusters
def find_statistics(edge_list_file, index_gene_file, reverse=True):
T = load_edge_list(edge_list_file)
index_to_gene, gene_to_index = load_index_gene(index_gene_file)
return compute_statistics(T, index_to_gene, reverse)
def run(args):
# Load data.
if args.verbose:
progress('Loading data...')
assert len(args.component_files) == len(args.index_gene_files) == len(
args.edge_list_files) == len(args.networks) == len(args.scores)
index_to_gene_collection = dict()
gene_to_index_collection = dict()
edge_list_collection = dict()
components_collection = dict()
for network_label, score_label, index_gene_file, edge_list_file, component_file in zip(
args.networks, args.scores, args.index_gene_files,
args.edge_list_files, args.component_files):
index_to_gene, gene_to_index = load_index_gene(index_gene_file)
edge_list = set(
frozenset(edge) for edge in load_edge_list(
edge_list_file, index_to_gene, unweighted=True))
components = load_components(component_file)
index_to_gene_collection[(network_label, score_label)] = index_to_gene
gene_to_index_collection[(network_label, score_label)] = gene_to_index
edge_list_collection[(network_label, score_label)] = edge_list
components_collection[(network_label, score_label)] = components
# Process data.
if args.verbose:
progress('Processing data...')
edge_to_networks = defaultdict(set)
edge_to_scores = defaultdict(set)
edge_to_pairs = defaultdict(set)
edge_to_tally = defaultdict(int)
for network_label, score_label in zip(args.networks, args.scores):
edge_list = edge_list_collection[(network_label, score_label)]
components = components_collection[(network_label, score_label)]
for component in components:
for u, v in combinations(component, 2):
edge = frozenset((u, v))
if edge in edge_list:
edge_to_tally[edge] += 1
thresholded_edges = set(edge for edge, tally in edge_to_tally.items()
if tally >= args.threshold)
G = nx.Graph()
G.add_edges_from(thresholded_edges)
consensus_results = sorted(sorted(
[sorted(x) for x in nx.connected_components(G)]),
key=len,
reverse=True)
# Save data.
if args.verbose:
progress('Saving data...')
if args.consensus_node_file is not None:
output_string = '\n'.join('\t'.join(x) for x in consensus_results)
with open(args.consensus_node_file, 'w') as f:
f.write(output_string)
if args.consensus_edge_file is not None:
output_string = '\n'.join(
'\t'.join(x) for x in sorted(map(sorted, thresholded_edges)))
with open(args.consensus_edge_file, 'w') as f:
f.write(output_string)
if args.verbose:
progress()