def test():
out_directory = "../test_128"
in_directory = "../test_trees_128"
image_directory = "../images/128"
with multiprocessing.Pool(processes=4) as pool:
parse_trees_to_json.parse_trees(in_directory)
vlmcs = VLMC.from_json_dir(in_directory)
lengths = [int(l) for l in np.logspace(2, 6, 10)]
print(lengths)
distances = calculate_distances_for_lengths(vlmcs, lengths,
out_directory,
image_directory, pool)
plot_results(vlmcs, distances, lengths, image_directory)
def distance_for_length(length, vlmcs, out_directory, image_directory, d):
repetitions = 5
distances = np.zeros(len(vlmcs))
for _ in range(repetitions):
train(vlmcs, length, out_directory)
parse_trees_to_json.parse_trees(out_directory)
new_vlmcs = VLMC.from_json_dir(out_directory)
pairs = pair_vlmcs(vlmcs, new_vlmcs)
# plot_vlmcs(pairs, image_directory)
rep_distance = distance_calculation(pairs, d)
distances += rep_distance / repetitions
print("{} done".format(length))
return distances
def test(args):
tree_directory = args.directory
out_directory = args.out_directory
parse_trees_to_json.parse_trees(tree_directory)
vlmcs = VLMC.from_json_dir(tree_directory)
cluster_class = parse_clustering_method(args)
d = parse_distance_method(args)
if args.name:
name = args.name
else:
name = cluster_class.__name__ + ", " + d.__class__.__name__
metrics = test_clustering(d, vlmcs, cluster_class)
try:
os.stat(out_directory)
except:
os.mkdir(out_directory)
plot_metrics(metrics, out_directory, name)
def test_distance_function(d,
tree_dir,
out_dir,
plot_distances=False,
plot_boxes=False):
parse_trees_to_json.parse_trees(tree_dir)
vlmcs = VLMC.from_json_dir(tree_dir)
metadata = get_metadata_for([vlmc.name for vlmc in vlmcs])
test_dir = tree_dir + "_test"
if os.path.isdir(test_dir):
parse_trees_to_json.parse_trees(test_dir)
test_vlmcs = VLMC.from_json_dir(test_dir)
else:
test_vlmcs = vlmcs
if out_dir is not None:
try:
os.stat(out_dir)
except:
os.mkdir(out_dir)
return test_distance_function_(d, vlmcs, test_vlmcs, metadata, out_dir,
True, False, plot_distances, plot_boxes)
parser.add_argument('--intersection', action='store_true')
parser.add_argument('--occurrence-probability-labels', action='store_true')
parser.add_argument(
'--directory',
type=str,
default='../trees_pst_better',
help='The directory which contains the vlmcs to be printed.')
parser.add_argument(
'--out-directory',
type=str,
default='../images',
help='The directory to where the images should be written.')
args = parser.parse_args()
try:
os.stat(args.out_directory)
except:
os.mkdir(args.out_directory)
parse_trees(args.directory, args.deltas)
vlmcs = VLMC.from_json_dir(args.directory)
metadata = get_metadata_for([vlmc.name for vlmc in vlmcs])
if args.intersection:
save_intersection(vlmcs, metadata, args.out_directory)
else:
save(vlmcs, metadata, args.out_directory, args.deltas,
args.occurrence_probability_labels)
def number_in_rank(metadata, key):
rank = Counter([v[key] for v in metadata.values()])
print("{}: size {}\n{}".format(key, len(rank.keys()), rank))
def order_analysis(vlmcs):
orders = [v.order for v in vlmcs]
order_counts = Counter(orders)
min_order = np.min(orders)
max_order = np.max(orders)
average_order = np.mean(orders)
print("Orders: {}, min: {}, max: {}, average: {}".format(
order_counts, min_order, max_order, average_order))
if __name__ == '__main__':
tree_dir = '../trees_virus_martin_all_96'
# tree_dir = '../trees_more_192'
parse_trees_to_json.parse_trees(tree_dir)
vlmcs = VLMC.from_json_dir(tree_dir)
metadata = get_metadata_for([vlmc.name for vlmc in vlmcs])
# vlmcs = [v for v in vlmcs if metadata[v.name]['genus'] == 'Ebolavirus']
# metadata = {k: v for k, v in metadata.items() if v['genus'] == 'Ebolavirus'}
# vlmcs = {metadata[v.name]['species']: v for v in vlmcs}
# vlmcs = [v for _, v in vlmcs.items()]
metadata = get_metadata_for([vlmc.name for vlmc in vlmcs])
number_in_ranks(metadata)
# order_analysis(vlmcs)
def get_vlmcs(out_directory, number_of_parameters):
directory = os.path.join(out_directory, str(number_of_parameters))
parse_trees(directory)
return VLMC.from_json_dir(directory)
def example_distance(tree_dir, image_dir):
parse_trees_to_json.parse_trees(tree_dir)
vlmcs = VLMC.from_json_dir(tree_dir)
d = FrobeniusNorm()
metadata = {v.name: {n: v.name for n in ['species', 'family', 'genus']} for v in vlmcs}
test_distance_function_(d, vlmcs, vlmcs, metadata, image_dir)
def parse_trees(args):
tree_dir = args.directory
parse_trees_to_json.parse_trees(tree_dir)
return VLMC.from_json_dir(tree_dir)