def build(method_name):
if method_name == 'sum':
return NamedFunctor("Sum Replicates",
lambda state, mode: aggregate_sum(state))
if method_name == 'mean':
return NamedFunctor("Average Replicates",
lambda state, mode: aggregate_mean(state))
if method_name == 'pick':
return NamedFunctor(
"Pick High Count Replicate",
lambda state, mode: aggregate_pick_high_count(state))
def build(meta_col_name: str, one_set: set, pair_strategy="paired_concat"):
if pair_strategy == "paired_concat":
return NamedFunctor(
"Target: " + meta_col_name + "(HouseholdConcat)", lambda state,
mode: matched_pair_concat(state, meta_col_name, one_set))
elif pair_strategy == "paired_subtract":
return NamedFunctor(
"Target: " + meta_col_name + "(HouseholdSubtract)", lambda state,
mode: matched_pair_subtract(state, meta_col_name, one_set))
elif pair_strategy == "unpaired":
return NamedFunctor(
"Target: " + meta_col_name,
lambda state, mode: _target(state, meta_col_name, one_set))
elif pair_strategy == "paired_subtract_sex_balanced":
return NamedFunctor(
"Target: " + meta_col_name,
lambda state, mode: matched_pair_subtract_sex_balanced(
state, meta_col_name, one_set))
def build_zebra(cov_thresh, cov_file):
zebra_df = pd.read_csv(cov_file)
zebra_pass_df = zebra_df[zebra_df.coverage_ratio > cov_thresh]
zebra_pass_list = zebra_pass_df['genome_id'].to_list()
def wrapped(state, mode):
state.df = state.df[state.df.columns.intersection(zebra_pass_list)]
return state.update_df(state.df)
return NamedFunctor("Zebra Filter:" + str(cov_thresh), wrapped)
def build(method_name, target_count=10000):
if method_name == 'none':
return NamedFunctor(
"No Normalization "
"(WARN: IGNORES COMPOSITIONALITY)", lambda state, mode: state)
if method_name == 'rarefy':
return NamedFunctor(
"Rarefy", lambda state, mode: rarefy_wrapper(state, target_count))
if method_name == 'divide_total':
return NamedFunctor(
"Truncate Ray To Simplex",
lambda state, mode: divide_total(state, target_count))
if method_name == 'ILR':
# Existing functions for this in skbio
raise NotImplemented()
if method_name == 'CLR':
return NamedFunctor("CLR Transform",
lambda state, mode: clr_wrapper(state))
if method_name == 'ALR':
# Existing functions for this in skbio
raise NotImplemented()
if method_name == 'AST':
raise NotImplemented()
def sum_columns():
return NamedFunctor("Sum All Columns",
lambda state, mode: _sum_columns(state))
def build_meta_encoder(col_name, encoder):
def wrapped(state, mode):
state.df[col_name] = state.meta_df[col_name].apply(encoder)
return state.update_df(state.df)
return NamedFunctor("Include " + col_name, wrapped)
def build_feature_set_transform(transformer):
def wrapped(state, mode):
return _apply_feature_transform(state, transformer)
return NamedFunctor(transformer.name, wrapped)
def build_prefix_filter(bad_prefixes):
return NamedFunctor(
"Filter Samples By ID Prefix",
lambda state, mode: _filter_out_sample_id_prefix(state, bad_prefixes))
def build_whitelist_metadata_value(metadata_column, metadata_values):
return NamedFunctor(
"Subset Samples by " + metadata_column + " for " +
str(metadata_values), lambda x, mode: _filter_by_metadata(
x, metadata_column, metadata_values))
def build_filter_out_empty_samples():
return NamedFunctor("Filter Empty Samples",
lambda state, mode: _filter_zero_sum(state))
def build(num_training_samples):
return NamedFunctor(
"Downsample Train Set: " + str(num_training_samples),
lambda state, mode: _downsample(state, mode, num_training_samples))
def build_exact_filter(bad_sample_ids):
return NamedFunctor(
"Remove Bad Samples",
lambda state, mode: _filter_out_sample_ids(state, bad_sample_ids))
def plot_correlation_matrix():
return NamedFunctor("Plot Correlation Heatmap",
lambda state, mode: _plot_correlation_matrix_deluxe(state))
def plot_scatter():
return NamedFunctor("Plot Scatter",
_plot_scatter)
def build_column_filter(chosen_columns):
return NamedFunctor(
"Restrict to columns (compositional): " + str(chosen_columns),
lambda state, mode: _restrict_columns_compositional(
state, chosen_columns))
def build_shared_filter():
return NamedFunctor("Filter Samples To Shared IDs",
lambda state, mode: _filter_by_shared_ids(state))
def build():
return NamedFunctor("Fix Sample IDs",
lambda state, mode: fix_sample_ids(state))