def _run_analysis(config_file, seed):
if seed is not None:
random.seed(seed)
config = paths.load_config(config_file)
alpha = config['concentration']['value']
if 'prior' in config['concentration']:
alpha_priors = config['concentration']['prior']
else:
alpha_priors = None
num_iters = config['num_iters']
density = config['density']
if density == 'pyclone_beta_binomial':
run_pyclone_beta_binomial_analysis(config_file, num_iters, alpha,
alpha_priors)
elif density == 'pyclone_binomial':
run_pyclone_binomial_analysis(config_file, num_iters, alpha,
alpha_priors)
else:
raise Exception(
'{0} is not a valid density for PyClone.'.format(density))
def _load_sample_data(file_name, error_rate, tumour_content):
'''
Load data from PyClone formatted input file.
'''
data = OrderedDict()
config = paths.load_config(file_name)
for mutation_dict in config['mutations']:
mutation = load_mutation_from_dict(mutation_dict)
data[mutation.id] = _get_pyclone_data(mutation, error_rate,
tumour_content)
return data
def load_table(config_file,
burnin=0,
min_size=0,
max_clusters=None,
mesh_size=101,
thin=1):
config = paths.load_config(config_file)
if config['density'] == 'pyclone_beta_binomial':
precision_file = paths.get_precision_trace_file(config_file)
precision = pd.read_csv(precision_file,
header=None,
compression='bz2',
sep='\t',
squeeze=True)
precision = precision.iloc[burnin::thin].mean()
density = PyCloneBetaBinomialDensity(GammaData(precision))
elif config['density'] == 'pyclone_binomial':
density = PyCloneBinomialDensity()
else:
raise Exception(
'Only pyclone_binomial and pyclone_beta_binomial density are supported.'
)
data, sample_ids = load_data(config_file)
labels = cluster_pyclone_trace(config_file,
burnin,
thin,
max_clusters=max_clusters)
labels = labels.set_index('mutation_id')
posteriors = []
for cluster_id, cluster_df in labels.groupby('cluster_id'):
mutation_ids = list(cluster_df.index)
cluster_data = [data[x] for x in mutation_ids]
for sample_id in sample_ids:
cluster_sample_data = [x[sample_id] for x in cluster_data]
cluster_sample_posterior = _compute_posterior(
cluster_sample_data, density, mesh_size)
cluster_sample_posterior['sample_id'] = sample_id
cluster_sample_posterior['cluster_id'] = cluster_id
cluster_sample_posterior['size'] = len(mutation_ids)
posteriors.append(cluster_sample_posterior)
df = pd.DataFrame(posteriors)
df = df.set_index(['sample_id', 'cluster_id', 'size'])
df = df.reset_index()
df = df[df['size'] >= min_size]
return df
def load_precision_params(config_file):
config = paths.load_config(config_file)
return config['beta_binomial_precision_params']
def load_init_method(config_file):
config = paths.load_config(config_file)
return config.get('init_method', 'disconnected')
def load_base_measure_params(config_file):
config = paths.load_config(config_file)
params = config['base_measure_params']
return params