def read_universal_counts(feature_names, state_names, file, file_type, feature_states_file):
""" This is a helper function to import global counts of each category of each feature,
which then define dirichlet distributions that are used as a prior for p_global.
Args:
feature_names(dict): the features names (internal and external)
state_names(dict): the category names (internal and external)
file (str): The file location of the pseudocounts for the prior
file_type (str): Two options:
´counts_file´: The counts are given in aggregate form (per features states)
´features_file´: Counts need to be extracted from a features file.
feature_states_file (str): The csv file containing the possible state values per feature.
Returns:
dict, str: The counts for each state in each feature and a log string.
"""
# Read the global counts from csv
if file_type == 'counts_file':
counts, feature_names_file, state_names_file = read_feature_occurrence_from_csv(file, feature_states_file)
else:
_, _, features, feature_names_file, state_names_file, *_ = read_features_from_csv(file, feature_states_file)
counts = np.sum(features, axis=0)
# # # Sanity checks # # #
# ´counts´ matrix has the right shape
n_features = len(feature_names_file['external'])
n_states = max(len(f_states) for f_states in state_names_file['external'])
assert counts.shape == (n_features, n_states)
# Are the data count data?
if not all(float(y).is_integer() for y in np.nditer(counts)):
out = f"The data in {file} must be count data."
raise ValueError(out)
# Same number of features?
assert len(feature_names['external']) == len(feature_names_file['external'])
assert len(feature_names['internal']) == len(feature_names_file['internal'])
# Same feature names?
for f in range(len(feature_names['external'])):
assert feature_names['external'][f] == feature_names_file['external'][f]
assert feature_names['internal'][f] == feature_names_file['internal'][f]
# Same number of categories?
assert len(state_names['external']) == len(state_names_file['external'])
assert len(state_names['internal']) == len(state_names_file['internal'])
# Same category names?
for f in range(len(state_names['external'])):
assert state_names['external'][f] == state_names_file['external'][f]
assert feature_names['internal'][f] == feature_names_file['internal'][f]
# Return with log message
log = f"Read universal counts from {file}"
return counts, log
def read_inheritance_counts(family_names, feature_names, state_names, files,
file_type, feature_states_file):
""" This is a helper function to import the counts of each feature in the families,
which define dirichlet distributions that are used as prior for p_family.
Args:
family_names(dict): the names of the families (internal and external)
feature_names(dict): the features names (internal and external)
state_names(dict): the category names (internal and external
files(dict): path to the file locations
file_type (str): Two options:
´counts_file´: The counts are given in aggregate form (per features states)
´features_file´: Counts need to be extracted from a features file.
feature_states_file (str): The csv file containing the possible state values per feature.
Returns:
dict, list: The dirichlet distribution per family for each category in each feature, and the categories
"""
n_families = len(family_names['external'])
n_features = len(feature_names['external'])
n_states = max([len(s) for s in state_names['external']])
counts_all = np.zeros([n_families, n_features, n_states])
log = str()
for fam_idx in range(n_families):
fam_name = family_names['external'][fam_idx]
if fam_name not in files:
log += f"No prior information for {fam_name}. Uniform prior used instead.\n"
continue
# Load counts for family ´fam_name´
file = files[fam_name]
if file_type == 'counts_file':
counts, feature_names_file, state_names_file = read_feature_occurrence_from_csv(
file, feature_states_file)
else:
_, _, features, feature_names_file, state_names_file, *_ = read_features_from_csv(
file, feature_states_file)
counts = np.sum(features, axis=0)
counts_all[fam_idx, :, :] = counts
log += f"Read counts for {fam_name} from {file}\n"
# # # Sanity checks # # #
if not all(float(y).is_integer() for y in np.nditer(counts)):
out = f"The data in {file} must be count data."
raise ValueError(out)
if len(feature_names['external']) != len(feature_names_file['external']) or \
len(feature_names['internal']) != len(feature_names_file['internal']):
out = "Different number of features in " + str(
file) + " as in features."
raise ValueError(out)
for f in range(0, len(feature_names['external'])):
if feature_names['external'][f] != feature_names_file['external'][
f]:
out = "The external feature " + str(f+1) + " in " + str(file) \
+ " differs from the one used in features."
raise ValueError(out)
if feature_names['internal'][f] != feature_names_file['internal'][
f]:
out = "The internal feature name " + str(f+1) + " in " + str(file) \
+ " differs from the one used in features."
raise ValueError(out)
if len(state_names['external']) != len(state_names_file['external']) or \
len(state_names['internal']) != len(state_names_file['internal']):
out = "Different number of features in " + str(
file) + " as in features."
raise ValueError(out)
for f in range(0, len(state_names['external'])):
if state_names['external'][f] != state_names_file['external'][f]:
out = "The external category names for feature " + str(f+1) + " in " + str(file) \
+ " differ from those used in features."
raise ValueError(out)
if feature_names['internal'][f] != feature_names_file['internal'][
f]:
out = "The internal category names for " + str(f+1) + " in " + str(file) \
+ " differ from those used in features."
raise ValueError(out)
return counts_all.astype(int), log