def test_group_human_scores(df_feats):
adata = adata_manipulations(df_feats)
df_small = make_small_dataset(adata)
df_gs = get_global_structure(use_cached=True)
df_small = df_small.merge(df_gs)
df = widen_df(df_small)
df = group_human_scores(df)
assert len(df) > 0
def test_make_regression_df(df_feats):
adata = adata_manipulations(df_feats)
df_small = make_small_dataset(adata)
df_gs = get_global_structure(use_cached=True)
df_small = df_small.merge(df_gs)
df = widen_df(df_small)
df = group_human_scores(df)
df = df.rename(rename_dict, axis="columns")
df, df_regression_info = make_regression_df(df)
assert len(df) > 0
assert len(df_regression_info) > 0
def load_data():
"""Monster function for loading and munging data for plots."""
# load main feature data
df_feats = load_main_feat_data()
# make anndata from feature data
# anndata as intermediate because our general purpose cleaning functions are written for anndata rather than pandas objects
adata = adata_manipulations(df_feats)
# make a df version of the feature data that only has a handful of simple features
df_small = make_small_dataset(adata)
# load in global structure features (DNN area classifier + radon transform stuff)
df_gs = get_global_structure()
# merge in the global structure metrics
df_small = df_small.merge(df_gs)
# make wide version
df = widen_df(df_small)
# group manual human structure scores into coarser bins
df = group_human_scores(df)
# clean up feauter/column names on the dataframes
df = df.rename(rename_dict, axis="columns")
# add regressed organizational score and grab regression info
df, df_regression = make_regression_df(df)
# create version of feature data where FISH probes are unpaired (makes facet plots easier)
df_tidy = tidy_df(df)
# clean up feature/column names on the dataframes
df = df.rename(rename_dict, axis="columns")
df_tidy = df_tidy.rename(rename_dict, axis="columns")
# clean up FISH probe names to drop amplifier ID
df, df_tidy = clean_probe_names(df, df_tidy)
# move from counts to count densities (normalize to cell area)
df, df_tidy = add_densities(df, df_tidy)
return df, df_tidy, df_regression