class AlonsClassifier:
def __init__(self, args, run=run, progress=progress):
self.run = run
self.progress = progress
A = lambda x: args.__dict__[x] if x in args.__dict__ else None
self.gene_coverages_data_file_path = A('data_file')
self.gene_detection_data_file_path = A('gene_detection_data_file')
self.profile_db_path = A('profile_db')
self.output_file_prefix = A('output_file_prefix')
self.alpha = A('alpha')
self.beta = A('beta')
self.gamma = A('gamma')
self.eta = A('eta')
self.zeta = A('zeta')
self.additional_layers_to_append = A('additional_layers_to_append')
self.samples_information_to_append = A('samples_information_to_append')
self.number_of_positive_samples = None
self.collection_name = A('collection_name')
self.bin_id = A('bin_id')
self.bin_ids_file_path = A('bin_ids_file')
self.store_gene_detections_and_gene_coverages_tables = A(
'store_gene_detections_and_gene_coverages_tables')
self.gene_coverages = {}
self.gene_detection = {}
self.samples = {}
self.positive_samples = {}
self.negative_samples = {}
self.gene_class_information = {}
self.samples_information = {}
self.profile_db = {}
self.sanity_check()
if self.profile_db_path is None:
self.get_data_from_txt_file()
else:
# load sample list and gene_coverage_dict from the merged profile db
args.init_gene_coverages = True
if self.collection_name:
self.summary = summarizer.ProfileSummarizer(args)
self.summary.init()
else:
self.profile_db = ProfileSuperclass(args)
self.profile_db.init_gene_coverages_and_detection_dicts()
self.gene_coverages = self.profile_db.gene_coverages_dict
self.gene_detection = self.profile_db.gene_detection_dict
self.samples = set(
next(iter(self.gene_coverages.values())).keys())
def sanity_check(self):
"""Basic sanity check for class inputs"""
if self.profile_db_path is None and self.gene_coverages_data_file_path is None:
raise ConfigError(
"You must provide either a profile.db or a gene coverage matrix data file"
)
if self.profile_db_path and self.gene_coverages_data_file_path:
raise ConfigError(
"You provided both a profile database and a gene coverage matrix data file, you \
must provide only one or the other (hint: if you have a profile database, the use that"
)
# checking output file
filesnpaths.is_output_file_writable(self.output_file_prefix +
'-additional-layers.txt',
ok_if_exists=False)
# checking alpha
if not isinstance(self.alpha, float):
raise ConfigError("alpha value must be a type float.")
if self.alpha <= 0 or self.alpha > 1:
raise ConfigError(
"alpha value must be greater than zero and a max of 1, the value you supplied %s"
% self.alpha)
# Checking beta
if not isinstance(self.beta, float):
raise ConfigError("beta value must be a type float.")
if self.beta <= 0:
raise ConfigError(
"beta value must be greater than zero, the value you supplied %s"
% self.beta)
# Checking gamma
if not isinstance(self.gamma, float):
raise ConfigError("Gamma value must be a type float.")
if self.gamma <= 0:
raise ConfigError(
"Gamma value must be greater than zero, the value you supplied %s"
% self.gamma)
# Checking eta
if self.eta <= 0 or self.eta > 1:
raise ConfigError(
"eta value must be greater than zero and a max of 1, the value you supplied %s"
% self.eta)
if self.collection_name:
if not self.profile_db_path:
raise ConfigError(
"You specified a collection name %s, but you provided a gene coverage matrix data file \
collections are only available when working with a profile database."
% self.collection_name)
def get_data_from_txt_file(self):
""" Reads the coverage data from TAB delimited file """
self.samples = set(
utils.get_columns_of_TAB_delim_file(
self.gene_coverages_data_file_path))
self.gene_coverages = utils.get_TAB_delimited_file_as_dictionary(
self.gene_coverages_data_file_path,
column_mapping=[int] + [float] * len(self.samples))
# checking if a gene_detection file was also supplied
if self.gene_detection_data_file_path:
self.gene_detection = utils.get_TAB_delimited_file_as_dictionary(
self.gene_detection_data_file_path,
column_mapping=[int] + [float] * len(self.samples))
# making sure that the tables are compatible, notice we're only checking if gene_detection contains everything that's in gene_coverages (and not vise versa)
for gene_id in self.gene_coverages:
if gene_id not in self.gene_detection:
raise ConfigError(
"Your tables are not compatible. For example gene_id %s is in %s, but not in %s"
% (gene_id, self.gene_coverages_data_file_path,
self.gene_detection_data_file_path))
gene_detection_sample_list = next(
iter(self.gene_detection.values())).keys()
for sample_id in next(iter(self.gene_coverages.values())).keys():
if sample_id not in gene_detection_sample_list:
raise ConfigError(
"Your tables are not compatible. For example sample_id %s is in %s, but not in %s"
% (sample_id, self.gene_coverages_data_file_path,
self.gene_detection_data_file_path))
def apply_func_to_genes_in_sample(self, func, list_of_genes=None):
""" Apply the give function on the list of genes in each sample. The function is expected to accept a list """
if not list_of_genes:
list_of_genes = self.gene_coverages.keys()
d = dict(
zip(self.samples, [
next(
map(func, [[
self.gene_coverages[gene_id][sample_id]
for gene_id in list_of_genes
]])) for sample_id in self.samples
]))
return d
def get_mean_coverage_in_samples(self, list_of_genes=None):
""" Returns a dictionary with of the average coverage value of the list of genes per sample. if no list of genes is
supplied then the average is calculated over all genes """
if not self.samples:
# if all samples don't contain the genome then return 0 for mean value
return 0
else:
mean_coverage_in_samples = self.apply_func_to_genes_in_sample(
np.mean, list_of_genes)
return mean_coverage_in_samples
def get_std_in_samples(self, list_of_genes=None):
""" Returns a dictionary with of the standard deviation of the coverage values of the list of genes per sample.
if no list of genes is supplied then the average is calculated over all genes """
std_in_samples = self.apply_func_to_genes_in_sample(
np.std, list_of_genes)
return std_in_samples
def get_detection_of_genes(self, mean_coverage_in_samples, std_in_samples):
""" Returns a dictionary (of dictionaries), where for each gene_id, and each sample_id the detection of the gene
is determined. The criteria for detection is having coverage that is greater than 0 and also that is not more
than gamma (default is gamma=3) standard deviations below the mean coverage in the sample.
Notice that the mean coverage isn't the mean of all genes in the sample necesarilly. In fact it would be the mean of
only the taxon-specific genes."""
detection_of_genes = {}
non_zero_non_detections = False
for gene_id in self.gene_coverages:
detection_of_genes[gene_id] = {}
detection_of_genes[gene_id]['number_of_detections'] = 0
detection_of_genes[gene_id][
'detected_in_non_positive_samples'] = False
for sample in self.samples:
if sample in self.positive_samples or sample not in self.negative_samples:
# getting gene detection according to coverage criteria
# samples that are ambiguous (neither negative nor positive)
detection_of_genes[gene_id][
sample] = self.gene_coverages[gene_id][sample] > max(
0, mean_coverage_in_samples[sample] -
self.gamma * std_in_samples[sample])
else:
# if the sample is a negative sample or the presence of the genome is unknown then the threshold for detections is
# more stringent and requires coverage significantly higher than the coverage of the TSC genes
detection_of_genes[gene_id][
sample] = self.gene_coverages[gene_id][sample] > max(
0, mean_coverage_in_samples[sample] +
self.gamma * std_in_samples[sample])
detection_of_genes[gene_id][
'detected_in_non_positive_samples'] += detection_of_genes[
gene_id][sample]
if self.gene_detection:
# if we have the gene detection (previously known as "percent covered") information then we will also use it to determine detection in samples:
gene_detection_above_threshold = self.gene_detection[
gene_id][sample] > self.zeta
detection_of_genes[gene_id][sample] = detection_of_genes[
gene_id][sample] * gene_detection_above_threshold
detection_of_genes[gene_id][
'number_of_detections'] += detection_of_genes[gene_id][
sample]
if self.gene_coverages[gene_id][sample] > 0 and self.gene_coverages[gene_id][sample] < mean_coverage_in_samples[sample] - \
self.gamma*std_in_samples[sample]:
non_zero_non_detections = True
if non_zero_non_detections:
# print('gene %s, in some sample has non-zero coverage %s, and it has been marked as not detected due '
# 'to the detection criteria' % (gene_id, data[gene_id][sample]))
print(
'some genes in some samples were marked as not detected due to the detection criteria'
)
return detection_of_genes
def get_samples_information(self,
detection_of_genes,
alpha,
genes_to_consider=None):
'''
Setting the values for the samples_information dictionary with the logical variable detection and the number of TSC genes
for each sample. In addition the total number of positive samples (i.e. samples that contain the reference genome) saved in
number_of_positive_samples.
'''
if not genes_to_consider:
# if no list of genes is supplied then considering all genes
genes_to_consider = detection_of_genes.keys()
samples_information = {}
self.number_of_positive_samples = 0
for sample_id in self.samples:
samples_information[sample_id] = {}
number_of_detected_genes_in_sample = len([
gene_id for gene_id in genes_to_consider
if detection_of_genes[gene_id][sample_id]
])
samples_information[sample_id][
'detection'] = number_of_detected_genes_in_sample > alpha * len(
genes_to_consider)
# There should be a better way to do this, but for now, if there is an intermediate value of genes that is
# detected then the sample would be considered neither positive nor negative
if number_of_detected_genes_in_sample > alpha * len(
genes_to_consider):
samples_information[sample_id]['detection'] = True
elif number_of_detected_genes_in_sample > 0.5 * alpha * len(
genes_to_consider):
samples_information[sample_id]['detection'] = None
else:
samples_information[sample_id]['detection'] = False
self.number_of_positive_samples += 1 if samples_information[
sample_id]['detection'] else 0
samples_information[sample_id][
'number_of_detected_genes'] = number_of_detected_genes_in_sample
self.samples_information = samples_information
self.run.warning('The number of positive samples is: %s ' %
self.number_of_positive_samples)
def get_adjusted_std_for_gene_id(self, gene_id, mean_coverage_in_samples,
detection_of_genes):
"""Returns the adjusted standard deviation for a gene_id. The adjusted standard deviation is calculated
only according to the positive samples (ones that contain the genome), and only in positive samples that
contain the gene (according to the gene detection criteria)."""
# Note: originally I thought I would only consider samples in which the genome was detected, but in fact,
# if a gene is detected in a sample in which the genome is not detected then that is a good sign that this is
# a TNS gene. But I still kept here the original definition of adjusted_std
# adjusted_std = np.std([d[gene_id, sample_id] / mean_coverage_in_samples[sample_id] for sample_id in samples if (
# detection_of_genes[gene_id][sample_id] and samples_information[sample_id])])
# First find all positive samples that also contain the gene
positive_samples_with_gene = []
for sample_id in self.positive_samples:
if detection_of_genes[gene_id][sample_id]:
positive_samples_with_gene.append(sample_id)
if not positive_samples_with_gene:
# If no positive samples contain the gene then we want the coverage consistency to be None
return -1, 0
else:
adjusted_gene_coverages = [
self.gene_coverages[gene_id][sample_id] /
mean_coverage_in_samples[sample_id]
for sample_id in positive_samples_with_gene
]
adjusted_mean = np.mean(adjusted_gene_coverages)
adjusted_std = np.std(adjusted_gene_coverages)
return adjusted_std, adjusted_mean
def get_adjusted_stds(self, mean_coverage_in_samples, detection_of_genes):
""" Returns a dictionary with the gene ids as keys and the adjusted standard deviation per gene"""
adjusted_std = {}
adjusted_mean = {}
# Make sure that there is at least one positive sample
if not self.number_of_positive_samples:
raise ConfigError(
"It seems that the reference genome is not detected in any of your samples. \
You might consider changing some of the parameters (for example --min-gene-detection, --min-portion-of-TSC \
or --core-min-detection).")
for gene_id in self.gene_coverages:
adjusted_std[gene_id], adjusted_mean[
gene_id] = self.get_adjusted_std_for_gene_id(
gene_id, mean_coverage_in_samples, detection_of_genes)
return adjusted_std, adjusted_mean
def get_gene_specificity(self, detection_of_genes):
""" Find all genes that are detected in """
gene_specificity = {}
for gene_id in detection_of_genes:
if detection_of_genes[gene_id]['detected_in_non_positive_samples']:
# if the gene is detected in at least one negative sample then gene_specificity is False
gene_specificity[gene_id] = False
elif detection_of_genes[gene_id]['number_of_detections'] == 0:
# if the gene is not detected in any sample then the gene_specificity is None
gene_specificity[gene_id] = None
else:
# if the gene is only detected in positive samples the gene_specificity is True
gene_specificity[gene_id] = True
return gene_specificity
def get_coverage_consistency(self, adjusted_stds, detection_of_genes,
beta):
"""For each gene if the adjusted standard deviation (to understand what this is refer to Alon Shaiber) is smaller
than beta then coverage_consistency is True, otherwise, coverage_consistency is False. If the gene is not
detected in positive samples then coverage consistency is not defined (get's a value 'None')"""
coverage_consistency = {}
for gene_id in adjusted_stds:
# if the gene is not detected in any sample then return None
if detection_of_genes[gene_id]['number_of_detections'] <= 1:
coverage_consistency[gene_id] = None
elif adjusted_stds[gene_id] == -1:
coverage_consistency[gene_id] = None
else:
if adjusted_stds[gene_id] < beta:
coverage_consistency[gene_id] = True
else:
coverage_consistency[gene_id] = False
return coverage_consistency
def get_taxon_specificity(self, coverage_consistency, gene_specificity):
taxon_specificity = {}
for gene_id in coverage_consistency:
if coverage_consistency[gene_id] is None or gene_specificity[
gene_id] is None:
taxon_specificity[gene_id] = None
elif coverage_consistency[gene_id] and gene_specificity[gene_id]:
taxon_specificity[gene_id] = 'TS'
else:
taxon_specificity[gene_id] = 'TNS'
return taxon_specificity
def get_loss_function_value(self, taxon_specificity, adjusted_stds, beta):
loss = 0
for gene_id in taxon_specificity:
if taxon_specificity[gene_id] == 'TS':
# Notice: here adjusted std includes the samples that don't have the genome detected in them (it kind of
# makes sense, because if the gene is detected even though the genome is not, then maybe it's not
# taxon-specific
loss += adjusted_stds[gene_id]
else:
loss += beta
return loss
def get_number_of_detections_for_gene(self, detection_of_genes, gene_id,
samples):
detections = 0
for sample_id in samples:
detections += detection_of_genes[gene_id][sample_id]
return detections
def get_core_accessory_info(self, detection_of_genes, gene_id, eta):
""" Returns 'core'/'accessory' classification for each gene. This is done using only the samples in which the
genome is detected """
if detection_of_genes[gene_id]['number_of_detections'] == 0:
return 'None'
elif self.get_number_of_detections_for_gene(
detection_of_genes, gene_id,
self.positive_samples) < eta * len(self.positive_samples):
return 'accessory'
else:
return 'core'
def get_gene_class(self, taxon_specificity, core_or_accessory):
if taxon_specificity == None or core_or_accessory == 'None':
return 'None'
elif taxon_specificity == 'TS':
if core_or_accessory == 'core':
return 'TSC'
elif core_or_accessory == 'accessory':
return 'TSA'
else:
print(
'%s is not valid. Value should be \'core\' or \'accessory\''
% core_or_accessory)
exit(1)
elif taxon_specificity == 'TNS':
if core_or_accessory == 'core':
return 'TNC'
elif core_or_accessory == 'accessory':
return 'TNA'
else:
print(
'%s is not valid. Value should be \'core\' or \'accessory\''
% core_or_accessory)
exit(1)
else:
print('%s is not valid. Value should be \'TS\' or \'TNS\'' %
taxon_specificity)
exit(1)
def report_gene_class_information(self):
C = lambda dictionary, field, value: len([
dict_id for dict_id in dictionary
if dictionary[dict_id][field] == value
])
for gene_class in ['TSC', 'TSA', 'TNC', 'TNA', 'None']:
self.run.info(
'Num class %s' % gene_class,
C(self.gene_class_information, 'gene_class', gene_class))
# TODO: report the number of negative samples and the number of NA samples
self.run.info('Num samples in which the genome is detected',
C(self.samples_information, 'detection', True),
mc='green')
def get_gene_classes(self):
""" returning the classification per gene along with detection in samples (i.e. for each sample, whether the
genome has been detected in the sample or not """
TSC_genes = set(self.gene_coverages.keys())
converged = False
loss = None
self.gene_class_information = {}
# Initializing all the samples to be positive
self.positive_samples = self.samples
self.negative_samples = {}
self.adjusted_stds = 0
self.adjusted_mean = dict.fromkeys(TSC_genes, 1)
while not converged:
# mean of coverage of all TS genes in each sample
mean_coverage_of_TS_in_samples = self.get_mean_coverage_in_samples(
TSC_genes)
# Get the standard deviation of the taxon-specific genes in a sample
# TODO: right now, single copy, and multi-copy genes would be treated identically. Hence, multi-copy genes
# would skew both the mean and the std of the taxon-specific genes.
std_of_TS_in_samples = self.get_std_in_samples(TSC_genes)
detection_of_genes = self.get_detection_of_genes(
mean_coverage_of_TS_in_samples, std_of_TS_in_samples)
self.get_samples_information(detection_of_genes, self.alpha,
TSC_genes)
self.positive_samples = {
sample_id
for sample_id in self.samples
if self.samples_information[sample_id]['detection']
}
self.negative_samples = {
sample_id
for sample_id in self.samples
if self.samples_information[sample_id]['detection'] == False
}
self.adjusted_stds, self.adjusted_mean = self.get_adjusted_stds(
mean_coverage_of_TS_in_samples, detection_of_genes)
coverage_consistency = self.get_coverage_consistency(
self.adjusted_stds, detection_of_genes, self.beta)
gene_specificity = self.get_gene_specificity(detection_of_genes)
taxon_specificity = self.get_taxon_specificity(
coverage_consistency, gene_specificity)
new_loss = self.get_loss_function_value(taxon_specificity,
self.adjusted_stds,
self.beta)
epsilon = 2 * self.beta
if loss is not None:
converged = True
loss = new_loss
self.run.warning('current value of loss function: %s ' % loss)
for gene_id in self.gene_coverages:
# setup a dict for gene id:
g = {}
g['gene_specificity'] = gene_specificity[gene_id]
g['gene_coverage_consistency'] = coverage_consistency[gene_id]
g['number_of_detections'] = detection_of_genes[gene_id][
'number_of_detections']
g['core_or_accessory'] = self.get_core_accessory_info(
detection_of_genes, gene_id, self.eta)
g['gene_class'] = self.get_gene_class(
taxon_specificity[gene_id], g['core_or_accessory'])
g['adjusted_stds'] = self.adjusted_stds[gene_id]
g['adjusted_mean'] = self.adjusted_mean[gene_id]
# counting the number of positive samples that contain the gene
g['detection_in_positive_samples'] = len([
sample_id for sample_id in self.positive_samples
if detection_of_genes[gene_id][sample_id]
])
# Getting the portion of positive samples that contain the gene
g['portion_detected'] = g[
'detection_in_positive_samples'] / len(
self.positive_samples
) if g['detection_in_positive_samples'] else 0
self.gene_class_information[gene_id] = g
TSC_genes = {
gene_id
for gene_id in self.gene_class_information
if self.gene_class_information[gene_id]['gene_class'] == 'TSC'
}
self.report_gene_class_information()
self.get_samples_information(detection_of_genes,
self.alpha,
genes_to_consider=TSC_genes)
def get_specificity_from_class_id(self, class_id):
try:
class_id = int(class_id)
except:
raise ConfigError(
"Classes must be of type integer. You sent this: ", class_id)
classes = {0: 'None', 1: 'TS', 2: 'TS', 3: 'TS', 4: 'TNS', 5: 'TNS'}
try:
return classes(class_id)
except:
raise ConfigError(
"The class id '%d' is not a valid one. Try one of these: '%s'"
% (class_id, ', '.join(list(classes.keys()))))
def save_gene_class_information_in_additional_layers(
self, additional_description=''):
if not self.additional_layers_to_append:
additional_column_titles = []
additional_layers_dict = self.gene_class_information
else:
additional_column_titles = utils.get_columns_of_TAB_delim_file(
self.additional_layers_to_append)
additional_layers_dict = utils.get_TAB_delimited_file_as_dictionary(
self.additional_layers_to_append,
dict_to_append=self.gene_class_information,
assign_none_for_missing=True,
column_mapping=[int] + [str] * len(additional_column_titles))
if additional_description:
additional_description = '-' + additional_description
additional_layers_file_name = self.output_file_prefix + additional_description + '-additional-layers.txt'
headers = headers = [
'gene_callers_id', 'gene_class', 'number_of_detections',
'portion_detected', 'gene_specificity',
'gene_coverage_consistency', 'core_or_accessory', 'adjusted_mean',
'adjusted_stds'
] + additional_column_titles
utils.store_dict_as_TAB_delimited_file(additional_layers_dict,
additional_layers_file_name,
headers=headers)
def save_samples_information(self, additional_description=''):
if not self.samples_information_to_append:
samples_information_column_titles = list(
self.samples_information[next(iter(self.samples_information))])
samples_information_dict = self.samples_information
else:
samples_information_column_titles = utils.get_columns_of_TAB_delim_file(
self.samples_information_to_append)
column_mapping = [str
] * (len(samples_information_column_titles) + 2)
self.run.warning(self.samples_information)
samples_information_dict = utils.get_TAB_delimited_file_as_dictionary(
self.samples_information_to_append,
dict_to_append=self.samples_information,
assign_none_for_missing=True,
column_mapping=column_mapping)
if additional_description:
additional_description = '-' + additional_description
samples_information_file_name = self.output_file_prefix + additional_description + '-samples-information.txt'
utils.store_dict_as_TAB_delimited_file(
samples_information_dict,
samples_information_file_name,
headers=['samples'] + samples_information_column_titles)
def save_gene_detection_and_coverage(self, additional_description=''):
if additional_description:
prefix = self.output_file_prefix + '-' + additional_description
else:
prefix = self.output_file_prefix
gene_coverages_file_name = prefix + '-gene-coverages.txt'
gene_detections_file_name = prefix + '-gene-detections.txt'
utils.store_dict_as_TAB_delimited_file(self.gene_coverages,
gene_coverages_file_name)
utils.store_dict_as_TAB_delimited_file(self.gene_detection,
gene_detections_file_name)
def get_coverage_and_detection_dict(self, bin_id):
_bin = summarizer.Bin(self.summary, bin_id)
self.gene_coverages = _bin.gene_coverages
self.gene_detection = _bin.gene_detection
self.samples = set(next(iter(self.gene_coverages.values())).keys())
def classify(self):
if self.collection_name:
bin_names_in_collection = self.summary.bin_ids
if self.bin_ids_file_path:
filesnpaths.is_file_exists(self.bin_ids_file_path)
bin_names_of_interest = [
line.strip()
for line in open(self.bin_ids_file_path).readlines()
]
missing_bins = [
b for b in bin_names_of_interest
if b not in bin_names_in_collection
]
if len(missing_bins):
raise ConfigError(
"Some bin names you declared do not appear to be in the collection %s. \
These are the bins that are missing: %s, these are the bins that are \
actually in your collection: %s" %
(self.collection_name, missing_bins,
bin_names_in_collection))
elif self.bin_id:
if self.bin_id not in bin_names_in_collection:
raise ConfigError("The bin you declared, %s, does not appear to be in the collection %s." \
% (self.bin_id, self.collection_name))
bin_names_of_interest = [self.bin_id]
else:
bin_names_of_interest = bin_names_in_collection
for bin_id in bin_names_of_interest:
self.run.info_single('Classifying genes in bin: %s' % bin_id)
self.get_coverage_and_detection_dict(bin_id)
self.get_gene_classes()
self.save_gene_class_information_in_additional_layers(bin_id)
self.save_samples_information(bin_id)
if self.store_gene_detections_and_gene_coverages_tables:
self.save_gene_detection_and_coverage(bin_id)
else:
# No collection provided so running on the entire detection table
self.get_gene_classes()
self.save_gene_class_information_in_additional_layers()
self.save_samples_information()
if self.store_gene_detections_and_gene_coverages_tables:
self.save_gene_detection_and_coverage()
class AlonsClassifier:
def __init__(self, args, run=run, progress=progress):
self.run = run
self.progress = progress
A = lambda x: args.__dict__[x] if x in args.__dict__ else None
self.gene_coverages_data_file_path = A('data_file')
self.gene_detections_data_file_path = A('gene_detection_data_file')
self.profile_db_path = A('profile_db')
self.output_file_prefix = A('output_file_prefix')
self.alpha = A('alpha')
self.beta = A('beta')
self.gamma = A('gamma')
self.eta = A('eta')
self.zeta = A('zeta')
self.additional_layers_to_append = A('additional_layers_to_append')
self.samples_information_to_append = A('samples_information_to_append')
self.collection_name = A('collection_name')
self.bin_id = A('bin_id')
self.bin_ids_file_path = A('bin_ids_file')
self.store_gene_detections_and_gene_coverages_tables = A(
'store_gene_detections_and_gene_coverages_tables')
self.exclude_samples = A('exclude_samples')
self.gene_coverages = pd.DataFrame.empty
self.gene_detections = pd.DataFrame.empty
self.samples = {}
self.positive_samples = pd.DataFrame.empty
self.number_of_positive_samples = None
self.negative_samples = pd.DataFrame.empty
self.number_of_negative_samples = None
self.gene_class_information = pd.DataFrame.empty
self.samples_information = pd.DataFrame.empty
self.profile_db = {}
self.gene_presence_absence_in_samples = pd.DataFrame.empty
self.gene_coverages_filtered = pd.DataFrame.empty
# check that there is a file like this
if self.exclude_samples:
filesnpaths.is_file_exists(self.exclude_samples)
self.samples_to_exclude = set([
l.split('\t')[0].strip()
for l in open(args.exclude_samples, 'rU').readlines()
])
run.info(
'Excluding Samples',
'The following samples will be excluded: %s' %
self.samples_to_exclude,
)
else:
self.samples_to_exclude = set([])
self.sanity_check()
if self.profile_db_path is None:
self.get_data_from_txt_file()
else:
# load sample list and gene_coverage_dict from the merged profile db
args.init_gene_coverages = True
if self.collection_name:
self.summary = summarizer.ProfileSummarizer(args)
self.summary.init()
else:
self.profile_db = ProfileSuperclass(args)
self.profile_db.init_gene_coverages_and_detection_dicts()
self.gene_coverages = pd.DataFrame.from_dict(
self.profile_db.gene_coverages_dict,
orient='index',
dtype=float)
self.gene_coverages.drop(self.samples_to_exclude,
axis=1,
inplace=True)
self.Ng = len(self.gene_coverages.index)
self.gene_detections = pd.DataFrame.from_dict(
self.profile_db.gene_detection_dict,
orient='index',
dtype=float)
self.gene_detections.drop(self.samples_to_exclude,
axis=1,
inplace=True)
self.samples = set(self.gene_coverages.columns)
def check_if_valid_portion_value(self, arg_name, arg_value):
""" Helper function to verify that an argument has a valid value for a non-zero portion (i.e. greater than zero and a max of 1)"""
if arg_value <= 0 or arg_value > 1:
raise ConfigError(
"%s value must be greater than zero and a max of 1, the value you supplied %s"
% (arg_name, arg_value))
def sanity_check(self):
"""Basic sanity check for class inputs"""
if self.profile_db_path is None and self.gene_coverages_data_file_path is None:
raise ConfigError(
"You must provide either a profile.db or a gene coverage self.gene_coverages_filtered data file"
)
if self.profile_db_path and self.gene_coverages_data_file_path:
raise ConfigError(
"You provided both a profile database and a gene coverage self.gene_coverages_filtered data file, you \
must provide only one or the other (hint: if you have a profile database, the use that"
)
# checking output file
filesnpaths.is_output_file_writable(self.output_file_prefix +
'-additional-layers.txt',
ok_if_exists=False)
# checking alpha
if not isinstance(self.alpha, float):
raise ConfigError("alpha value must be a type float.")
self.check_if_valid_portion_value("alpha", self.alpha)
# Checking beta
if not isinstance(self.beta, float):
raise ConfigError("beta value must be a type float.")
self.check_if_valid_portion_value("beta", self.beta)
if self.beta > self.alpha:
raise ConfigError(
"beta value must be smaller than alpha value. The beta value you specified is %s while the alpha value\
is %s" % (self.beta, self.alpha))
# Checking gamma
if not isinstance(self.gamma, float):
raise ConfigError("Gamma value must be a type float.")
self.check_if_valid_portion_value("gamma", self.gamma)
# Checking eta
self.check_if_valid_portion_value("eta", self.eta)
if self.collection_name:
if not self.profile_db_path:
raise ConfigError(
"You specified a collection name %s, but you provided a gene coverage self.gene_coverages_filtered data file \
collections are only available when working with a profile database."
% self.collection_name)
def get_data_from_txt_file(self):
""" Reads the coverage data from TAB delimited file """
self.gene_coverages = pd.read_table(self.gene_coverages_data_file_path,
sep='\t',
header=0,
index_col=0)
self.gene_coverages.drop(self.samples_to_exclude, axis=1, inplace=True)
self.Ng = len(self.gene_coverages.index)
self.samples = set(self.gene_coverages.columns.values)
# checking if a gene_detection file was also supplied
if self.gene_detections_data_file_path:
self.gene_detections = pd.read_table(
self.gene_coverages_data_file_path,
sep='\t',
header=0,
index_col=0)
self.gene_detections.drop(self.samples_to_exclude,
axis=1,
inplace=True)
# making sure that the tables are compatible, notice we're only checking if gene_detection contains everything that's in gene_coverages (and not vise versa)
for gene_id in self.gene_coverages.index:
if gene_id not in self.gene_detections.index:
raise ConfigError(
"Your tables are not compatible. For example gene_id %s is in %s, but not in %s"
% (gene_id, self.gene_coverages_data_file_path,
self.gene_detections_data_file_path))
for sample_id in self.samples:
if sample_id not in self.gene_detections.columns.values:
raise ConfigError(
"Your tables are not compatible. For example sample_id %s is in %s, but not in %s"
% (sample_id, self.gene_coverages_data_file_path,
self.gene_detections_data_file_path))
def set_gene_presence_absence_in_samples(self):
""" Determines the presence/absense of genes according to the gene detection threshold.
The following arguments are populated:
self.gene_presence_absence_in_samples
self.gene_coverages_filtered
"""
self.gene_presence_absence_in_samples = self.gene_detections > self.zeta
self.gene_coverages_filtered = self.gene_coverages.copy()
self.gene_coverages_filtered[~self.
gene_presence_absence_in_samples] = 0
def set_sample_detection_information(self):
""" Using the --genome-presence-threshold and the --genome-absence-threhold the samples are devided to three groups:
positive samples: samples in which the number of genes that are present (according to the --min-gene-detection threshold) is
greater than --genome-presence-threshold
negative samples: samples in which the number of genes that are present (according to the --min-gene-detection threshold) is
smaller than --genome-absence-threhold
ambiguous samples: all other samples (i.e. samples in which the number of genes that are present is between the two thresholds)
This function populates the following arguments of self:
self.positive_samples - a set of the positive sample ids self.negative_samples - a set of the negative sample ids
self.samples_information - dictionary with True, False, None for positive, negative and ambiguous samples respectively
"""
MDG_samples_information_table_name = 'MDG_classifier_samples_information'
MDG_samples_information_table_structure = [
'samples', 'presence', 'number_of_detected_genes',
'number_of_taxon_specific_core_detected'
]
MDG_samples_information_table_types = ['str', 'bool', 'int', 'int']
# create an empty dataframe
samples_information = pd.DataFrame(
index=self.samples,
columns=MDG_samples_information_table_structure[1:])
# Compute the number of detected genes per samples
number_of_genes_in_sampels = self.gene_presence_absence_in_samples.sum(
axis=0)
for sample in self.samples:
samples_information['number_of_detected_genes'][
sample] = number_of_genes_in_sampels[sample]
self.positive_samples = set(
number_of_genes_in_sampels[number_of_genes_in_sampels /
self.Ng > self.alpha].index)
self.negative_samples = set(number_of_genes_in_sampels[
number_of_genes_in_sampels /
self.Ng > self.beta].index) - self.positive_samples
for sample in self.positive_samples:
samples_information['presence'][sample] = True
for sample in self.negative_samples:
samples_information['presence'][sample] = False
self.samples_information = samples_information
self.number_of_positive_samples = len(self.positive_samples)
self.number_of_negative_samples = len(self.negative_samples)
self.run.warning('The number of positive samples is: %d ' %
self.number_of_positive_samples)
self.run.warning('The number of negative samples is: %d ' %
self.number_of_negative_samples)
self.run.warning('The number of ambiuous samples is: %d ' %
(len(self.samples - self.negative_samples -
self.positive_samples)))
def get_taxon_specific_genes_in_samples(self, additional_description=''):
""" Use only positive samples to identify the single copy core genes:
Assumption: At least 25% of the genes are single copy and taxon specific (so that it is included in the Q1Q3 range).
a. Sort coverage vaules
b. Sliding window (with 50% overlapping windows) that has least outliers with the smallest range. The reason for 50% \
overlap is that this guarentees that if the assumption that at least 25% are single copy taxon specific then, it is \
guarentees that at least one window will have Q1Q3 composed only from taxon specific genes.
i.Run sliding window and for each step calculate the number of outliers and the range
ii.Sort according the number of outliers and then according to the
"""
# get the indixes that sort each column separately TODO: delete this part if you don't need it
sorting_indexes = self.gene_coverages_filtered.apply(
lambda x: x.sort_values(ascending=True).index)
# TODO: I'm creating a copy but probably I don't need to
coverages_pdf_output = self.output_file_prefix + additional_description + '-coverages.pdf'
coverages_pdf_output_file = PdfPages(coverages_pdf_output)
# creating dataframe for non-outliers (see usage below)
non_outliers = pd.DataFrame().reindex_like(self.gene_coverages)
sorting_indexes = self.gene_coverages_filtered.apply(
lambda x: x.sort_values(ascending=True).index)
print(self.samples_information)
print(self.positive_samples)
print(self.negative_samples)
print(self.samples)
for sample in self.positive_samples:
# a vector of the coverages for the sample
v = self.gene_coverages_filtered[sample].copy()
# set coverage to zero if not above gene detection threshold
v[self.gene_detections[sample] < self.zeta] = 0
# get non-outliers according to the Interquartile range
non_outliers[sample] = get_non_outliers(v)
plot_outliers(coverages_pdf_output_file, v, non_outliers[sample],
sample)
coverages_pdf_output_file.close()
# finding the genes that are non-outliers in the majority of samples
# the required majority is defined by the user defined argument self.eta (--core-min-detection)
non_outliers_all = non_outliers[non_outliers.sum(
axis=1) >= self.number_of_positive_samples * self.eta].index
return non_outliers_all
def get_gene_specificity(self, detection_of_genes):
""" Find all genes that are detected in """
gene_specificity = {}
for gene_id in detection_of_genes:
if detection_of_genes[gene_id]['detected_in_non_positive_samples']:
# if the gene is detected in at least one negative sample then gene_specificity is False
gene_specificity[gene_id] = False
elif detection_of_genes[gene_id]['number_of_detections'] == 0:
# if the gene is not detected in any sample then the gene_specificity is None
gene_specificity[gene_id] = None
else:
# if the gene is only detected in positive samples the gene_specificity is True
gene_specificity[gene_id] = True
return gene_specificity
def get_coverage_consistency(self, adjusted_stds, detection_of_genes,
beta):
"""For each gene if the adjusted standard deviation (to understand what this is refer to Alon Shaiber) is smaller
than beta then coverage_consistency is True, otherwise, coverage_consistency is False. If the gene is not
detected in positive samples then coverage consistency is not defined (get's a value 'None')"""
coverage_consistency = {}
for gene_id in adjusted_stds:
# if the gene is not detected in any sample then return None
if detection_of_genes[gene_id]['number_of_detections'] <= 1:
coverage_consistency[gene_id] = None
elif adjusted_stds[gene_id] == -1:
coverage_consistency[gene_id] = None
else:
if adjusted_stds[gene_id] < beta:
coverage_consistency[gene_id] = True
else:
coverage_consistency[gene_id] = False
return coverage_consistency
def get_taxon_specificity(self, coverage_consistency, gene_specificity):
taxon_specificity = {}
for gene_id in coverage_consistency:
if coverage_consistency[gene_id] is None or gene_specificity[
gene_id] is None:
taxon_specificity[gene_id] = None
elif coverage_consistency[gene_id] and gene_specificity[gene_id]:
taxon_specificity[gene_id] = 'TS'
else:
taxon_specificity[gene_id] = 'TNS'
return taxon_specificity
def get_number_of_detections_for_gene(self, detection_of_genes, gene_id,
samples):
detections = 0
for sample_id in samples:
detections += detection_of_genes[gene_id][sample_id]
return detections
def get_core_accessory_info(self, detection_of_genes, gene_id, eta):
""" Returns 'core'/'accessory' classification for each gene. This is done using only the samples in which the
genome is detected """
if detection_of_genes[gene_id]['number_of_detections'] == 0:
return 'None'
elif self.get_number_of_detections_for_gene(
detection_of_genes, gene_id,
self.positive_samples) < eta * len(self.positive_samples):
return 'accessory'
else:
return 'core'
def get_gene_class(self, taxon_specificity, core_or_accessory):
if taxon_specificity == None or core_or_accessory == 'None':
return 'None'
elif taxon_specificity == 'TS':
if core_or_accessory == 'core':
return 'TSC'
elif core_or_accessory == 'accessory':
return 'TSA'
else:
print(
'%s is not valid. Value should be \'core\' or \'accessory\''
% core_or_accessory)
exit(1)
elif taxon_specificity == 'TNS':
if core_or_accessory == 'core':
return 'TNC'
elif core_or_accessory == 'accessory':
return 'TNA'
else:
print(
'%s is not valid. Value should be \'core\' or \'accessory\''
% core_or_accessory)
exit(1)
else:
print('%s is not valid. Value should be \'TS\' or \'TNS\'' %
taxon_specificity)
exit(1)
def report_gene_class_information(self):
# TODO: change to pandas
C = lambda dictionary, field, value: len([
dict_id for dict_id in dictionary
if dictionary[dict_id][field] == value
])
for gene_class in ['TSC', 'TSA', 'TNC', 'TNA', 'None']:
self.run.info(
'Num class %s' % gene_class,
C(self.gene_class_information, 'gene_class', gene_class))
# TODO: report the number of negative samples and the number of NA samples
self.run.info('Num samples in which the genome is detected',
C(self.samples_information, 'detection', True),
mc='green')
def get_gene_classes(self):
""" returning the classification per gene along with detection in samples (i.e. for each sample, whether the
genome has been detected in the sample or not """
# need to start a new gene_class_information dict
# this is due to the fact that if the algorithm is ran on a list of bins then this necessary
self.gene_class_information = pd.DataFrame(
index=self.gene_coverages.index, columns=['gene_class'])
# use a gene detection threshold to determine gene presence/absence in samples
self.set_gene_presence_absence_in_samples()
# use a gene presence threshold classify samples as positive, negative or ambiguous
self.set_sample_detection_information()
#
non_outliers_all = self.get_taxon_specific_genes_in_samples()
# set the gene classes
for gene_id in self.gene_coverages_filtered.index:
if gene_id in non_outliers_all:
self.gene_class_information['gene_class'][
gene_id] = self.get_gene_class('TS', 'core')
else:
self.gene_class_information['gene_class'][
gene_id] = self.get_gene_class(None, None)
def get_specificity_from_class_id(self, class_id):
try:
class_id = int(class_id)
except:
raise ConfigError(
"Classes must be of type integer. You sent this: ", class_id)
classes = {0: 'None', 1: 'TS', 2: 'TS', 3: 'TS', 4: 'TNS', 5: 'TNS'}
try:
return classes(class_id)
except:
raise ConfigError(
"The class id '%d' is not a valid one. Try one of these: '%s'"
% (class_id, ', '.join(list(classes.keys()))))
def save_gene_class_information_in_additional_layers(
self, additional_description=''):
if not self.additional_layers_to_append:
additional_column_titles = []
additional_layers_df = self.gene_class_information
else:
additional_layers_df = pd.read_table(
self.additional_layers_to_append)
try:
# concatinating the gene_class_information with the user provided additional layers
additional_layers_df.join(self.gene_class_information,
how='outer')
except ValueError as e:
raise ConfigError(
"Something went wrong. This is what we know: %s. This could be happening because \
you have columns in your --additional-layers file with the following title: %s"
% (e, self.gene_class_information.columns.tolist()))
if additional_description:
additional_description = '-' + additional_description
additional_layers_file_name = self.output_file_prefix + additional_description + '-additional-layers.txt'
additional_layers_df.to_csv(additional_layers_file_name,
sep='\t',
index_label='gene_callers_id')
def save_samples_information(self, additional_description=''):
# TODO: there used to be this here:
#self.run.warning(self.samples_information)
if not self.samples_information_to_append:
samples_information_df = self.samples_information
else:
samples_information_df = pd.read_table(
self.samples_information_to_append)
try:
# concatinating the samples_information with the user provided samples_information file
samples_information_df.join(self.samples_information,
how='outer')
except ValueError as e:
raise ConfigError(
"Something went wrong. This is what we know: %s. This could be happening because \
you have columns in your --additional-layers file with the following title: %s"
% (e, self.samples_information.columns.tolist()))
if additional_description:
additional_description = '-' + additional_description
samples_information_file_name = self.output_file_prefix + additional_description + '-samples-information.txt'
samples_information_df.to_csv(samples_information_file_name,
sep='\t',
index_label='samples')
def save_gene_detection_and_coverage(self, additional_description=''):
if additional_description:
prefix = self.output_file_prefix + '-' + additional_description
else:
prefix = self.output_file_prefix
gene_coverages_file_name = prefix + '-gene-coverages.txt'
gene_detections_file_name = prefix + '-gene-detections.txt'
self.gene_coverages.to_csv(gene_coverages_file_name,
sep='\t',
index_label='gene_callers_id')
self.gene_detections.to_csv(gene_detections_file_name,
sep='\t',
index_label='gene_callers_id')
def get_coverage_and_detection_dict(self, bin_id):
_bin = summarizer.Bin(self.summary, bin_id)
self.gene_coverages = pd.DataFrame.from_dict(_bin.gene_coverages,
orient='index',
dtype=float)
print(self.gene_coverages)
self.gene_coverages.drop(self.samples_to_exclude, axis=1, inplace=True)
self.Ng = len(self.gene_coverages.index)
self.gene_detections = pd.DataFrame.from_dict(_bin.gene_detection,
orient='index',
dtype=float)
self.gene_detections.drop(self.samples_to_exclude,
axis=1,
inplace=True)
self.samples = set(self.gene_coverages.columns.values)
def classify(self):
if self.collection_name:
bin_names_in_collection = self.summary.bin_ids
if self.bin_ids_file_path:
filesnpaths.is_file_exists(self.bin_ids_file_path)
bin_names_of_interest = [
line.strip()
for line in open(self.bin_ids_file_path).readlines()
]
missing_bins = [
b for b in bin_names_of_interest
if b not in bin_names_in_collection
]
if len(missing_bins):
raise ConfigError(
"Some bin names you declared do not appear to be in the collection %s. \
These are the bins that are missing: %s, these are the bins that are \
actually in your collection: %s" %
(self.collection_name, missing_bins,
bin_names_in_collection))
elif self.bin_id:
if self.bin_id not in bin_names_in_collection:
raise ConfigError("The bin you declared, %s, does not appear to be in the collection %s." \
% (self.bin_id, self.collection_name))
bin_names_of_interest = [self.bin_id]
else:
bin_names_of_interest = bin_names_in_collection
for bin_id in bin_names_of_interest:
self.run.info_single('Classifying genes in bin: %s' % bin_id)
self.get_coverage_and_detection_dict(bin_id)
self.get_gene_classes()
self.save_gene_class_information_in_additional_layers(bin_id)
self.save_samples_information(bin_id)
if self.store_gene_detections_and_gene_coverages_tables:
self.save_gene_detection_and_coverage(bin_id)
else:
# No collection provided so running on the entire detection table
self.get_gene_classes()
self.save_gene_class_information_in_additional_layers()
self.save_samples_information()
if self.store_gene_detections_and_gene_coverages_tables:
self.save_gene_detection_and_coverage()
