def test_generate_sizes(self):
"""Checks whether the set sizes are correctly returned. """
perm = 10
nperm = 10
random, degree = generate_null(networks, core=2, n=perm, npos=10)
results = generate_sizes(networks,
random_models=random,
core=2,
degree_models=degree,
prev=None,
fractions=False,
perm=nperm,
sizes=[1],
sign=True)
# 126: 2 set operations * 21 networks * 3 groups
self.assertEqual(len(results), 126)
def test_compare_set_sizes(self):
"""
Given a pandas dataframe with set sizes across groups of networks,
this function should return a dataframe with statistics on these set sizes.
"""
set_values = generate_sizes(networks,
random,
degree,
fractions=None,
prev=None,
core=2,
sign=True,
perm=10,
sizes=[0.6, 1])
results = compare_set_sizes(set_values)
results = results[results['Comparison'] == 'Random']
results = results[results['Measure'] == 'Intersection 0.6']
self.assertGreater(0.1, results['P'].iloc[0])
def test_generate_stat_rows(self):
"""
Given a pandas dataframe with results, this function should add a row.
"""
set_values = generate_sizes(networks,
random,
degree,
fractions=None,
prev=None,
core=2,
sign=True,
perm=10,
sizes=[0.6, 1])
results = compare_set_sizes(set_values)
new_results = _generate_stat_rows(results,
group='b',
comparison='test',
operation='test',
p='0.05',
ptype='test',
node=None)
self.assertGreater(len(new_results), len(results))
def test_generate_size_differences(self):
"""
Tests whether the size differences are returned correctly.
:return:
"""
perm = 10
nperm = 10
new = {'a': [networks['a'][0], networks['a'][0], networks['b'][0]]}
random, degree = generate_null(new, core=2, n=perm, npos=10)
results = generate_sizes(new,
random_models=random,
core=2,
degree_models=degree,
prev=None,
fractions=False,
perm=nperm,
sizes=[0.6, 1],
sign=True)
results = generate_size_differences(results, sizes=[0.6, 1])
results = results[results['Network'] == 'Input']
results = results[results['Interval'] == '1->1']
self.assertEqual(results['Set size'].iloc[0], 4.0)
def model_calcs(networks, args):
"""
Function for generating null models and carrying out calculations.
:param networks: Dictionary with folder name as key and values as tuples (name, network object).
:param args: Settings for running anuran
:return:
"""
if args['core'] < 1:
args['core'] = 1
logger.info("Setting cores for multiprocessing to 1.")
# export intersections
for size in args['size']:
for group in networks:
shared_edges = _intersection(networks[group], float(size), sign=args['sign'], edgelist=True)
g = _construct_intersection(networks[group], shared_edges)
nx.write_graphml(g, args['fp'] + '_' + group + '_' + str(size) + '_intersection.graphml')
# first generate null models
try:
random, degree = generate_null(networks, n=args['perm'], npos=args['gperm'], core=args['core'], fraction=args['cs'],
prev=args['prev'])
except Exception:
logger.error('Could not generate null models!', exc_info=True)
sys.exit()
set_sizes = None
try:
set_sizes = generate_sizes(networks, random, degree, core=args['core'],
sign=args['sign'],
fractions=args['cs'], prev=args['prev'],
perm=args['nperm'], sizes=args['size'])
set_sizes.to_csv(args['fp'] + '_sets.csv')
set_differences = generate_size_differences(set_sizes, sizes=args['size'])
set_differences.to_csv(args['fp'] + '_set_differences.csv')
logger.info('Set sizes exported to: ' + args['fp'] + '_sets.csv')
except Exception:
logger.error('Failed to calculate set sizes!', exc_info=True)
sys.exit()
centralities = None
if args['centrality']:
try:
centralities = generate_ci_frame(networks, random, degree,
fractions=args['cs'], prev=args['prev'],
perm=args['nperm'], core=args['core'])
centralities.to_csv(args['fp'] + '_centralities.csv')
logger.info('Centralities exported to: ' + args['fp'] + '_centralities.csv')
except Exception:
logger.error('Could not rank centralities!', exc_info=True)
sys.exit()
if args['network']:
try:
graph_properties = generate_graph_frame(networks, random, degree,
fractions=args['cs'], core=args['prev'],
perm=args['nperm'])
graph_properties.to_csv(args['fp'] + '_graph_properties.csv')
logger.info('Graph properties exported to: ' + args['fp'] + '_graph_properties.csv')
except Exception:
logger.error('Could not estimate graph properties!', exc_info=True)
sys.exit()
samples = None
if args['sample']:
try:
samples = generate_sample_sizes(networks, random, degree,
sign=args['sign'], core=args['core'],
fractions=args['cs'], perm=args['nperm'], prev=args['prev'],
sizes=args['size'], limit=args['sample'], number=args['number'])
samples.to_csv(args['fp'] + '_subsampled_sets.csv')
logger.info('Subsampled set sizes exported to: ' + args['fp'] + '_subsampled_sets.csv')
except Exception:
logger.error('Failed to subsample networks!', exc_info=True)
sys.exit()
central_stats = None
if args['stats']:
if args['stats'] == 'True':
args['stats'] = True
# add code for pvalue estimation
set_stats = compare_set_sizes(set_sizes)
set_stats.to_csv(args['fp'] + '_set_stats.csv')
difference_stats = compare_set_sizes(set_differences)
difference_stats.to_csv(args['fp'] + '_difference_stats.csv')
if args['centrality'] and centralities is not None:
central_stats = compare_centralities(centralities, mc=args['stats'])
central_stats.to_csv(args['fp'] + '_centrality_stats.csv')
if args['network']:
graph_stats = compare_graph_properties(graph_properties)
graph_stats.to_csv(args['fp'] + '_graph_stats.csv')
# check if there is an order in the filenames
for group in networks:
prefixes = [x[0].split('_')[0] for x in networks[group]]
try:
prefixes = [int(x) for x in prefixes]
except ValueError:
pass
if all(isinstance(x, int) for x in prefixes):
centrality_correlation = correlate_centralities(group, centralities, mc=args['stats'])
centrality_correlation.to_csv(args['fp'] + '_centrality_correlation.csv')
graph_correlation = correlate_graph_properties(group, graph_properties)
graph_correlation.to_csv(args['fp'] + '_centrality_correlation.csv')
if args['draw']:
try:
for x in networks:
subset_sizes = set_sizes[set_sizes['Group'] == x]
draw_sets(subset_sizes, args['fp'] + '_' + x)
subset_differences = set_differences[set_differences['Group'] == x]
draw_set_differences(subset_differences, args['fp'] + '_' + x)
if args['centrality']:
subset_centralities = centralities[centralities['Group'] == x]
draw_centralities(subset_centralities, args['fp'] + '_' + x)
if args['sample']:
subset_samples = samples[samples['Group'] == x]
draw_samples(subset_samples, args['fp'] + '_' + x)
if args['network']:
subset_graphs = graph_properties[graph_properties['Group'] == x]
draw_graphs(subset_graphs, args['fp'] + '_' + x)
except Exception:
logger.error('Could not draw data!', exc_info=True)
sys.exit()
if central_stats is not None:
return central_stats