Beta diversity tweaks

CHANGELOG.md

@@ -18,6 +18,8 @@
 #### analysis
 ##### Fixed 🐛
 - Fixed column names slipping into column values when `motifs = True` combined with `transform = ALR` in `get_pca` (e802da1)
+##### Changed 🔄
+- Added distance matrix to beta diversity output
 
 #### draw
 ##### Changed 🔄

glycowork/motif/analysis.py

@@ -7,6 +7,8 @@
 import networkx as nx
 import statsmodels.api as sm
 import matplotlib.pyplot as plt
+from numpy import ndarray
+
 plt.rcParams.update({
     'font.size': 11, 'axes.labelsize': 12, 'axes.titlesize': 13,
     'xtick.labelsize': 10, 'ytick.labelsize': 10, 'axes.linewidth': 0.8,
@@ -1015,13 +1017,14 @@
     gamma: float = 0.1, # Uncertainty parameter for CLR transform
     custom_scale: float | dict = 0, # Ratio of total signal in group2/group1 for an informed scale model (or group_idx: mean(group)/min(mean(groups)) signal dict for multivariate)
     random_state: int | np.random.Generator | None = None # optional random state for reproducibility
-    ) -> pd.DataFrame: # DataFrame with diversity indices and test statistics
+    ) -> tuple[pd.DataFrame, pd.DataFrame ]: # First dataFrame with diversity indices and test statistics, second with beta-diversity distance matrix
   "Calculates alpha (Shannon/Simpson) and beta (ANOSIM/PERMANOVA) diversity measures from glycomics data"
   experiment = "diff" if group2 else "anova"
   df, df_org, group1, group2 = preprocess_data(df, group1, group2, experiment = experiment, motifs = motifs, impute = False,
                                                transform = transform, feature_set = feature_set, paired = paired, gamma = gamma,
                                                custom_scale = custom_scale, custom_motifs = custom_motifs, random_state = random_state)
   shopping_cart = []
+  distance_matrix = pd.DataFrame()
   group_sizes = group1 if not group2 else len(group1)*[1]+len(group2)*[2]
   group_counts = Counter(group_sizes)
   # Sample-size aware alpha via Bayesian-Adaptive Alpha Adjustment
@@ -1066,8 +1069,8 @@
         bc_diversity[index_1, index_2] = bc_pair
     b_df_out = pd.DataFrame.from_dict(bc_diversity, orient = 'index')
     out_len = int(np.sqrt(len(b_df_out)))
-    b_df_out_values = b_df_out.values.reshape(out_len, out_len)
-    beta_df_out = pd.DataFrame(data = b_df_out_values, index = range(out_len), columns = range(out_len))
+    distance_matrix = b_df_out.values.reshape(out_len, out_len)
+    beta_df_out = pd.DataFrame(data = distance_matrix, index = range(out_len), columns = range(out_len))
     if all(count > 1 for count in group_counts.values()):
       r, p = anosim(beta_df_out, group_sizes, permutations)
       b_test_stats = pd.DataFrame({'Metric': 'Beta diversity (ANOSIM)', 'p-val': p, 'Effect size': r}, index = [0])
@@ -1079,7 +1082,7 @@
   corrpvals, significance = correct_multiple_testing(df_out['p-val'], alpha)
   df_out["corr p-val"] = corrpvals
   df_out["significant"] = significance
-  return df_out.sort_values(by = 'p-val').sort_values(by = 'corr p-val').reset_index(drop = True)
+  return df_out.sort_values(by = 'p-val').sort_values(by = 'corr p-val').reset_index(drop = True), distance_matrix
 
 
 def get_SparCC(

tests/test_core_functions.py

@@ -4006,7 +4006,7 @@ def generate_group_data(base_proportions, n_samples=3, noise_scale=0.15):
     # Run biodiversity analysis
     results = get_biodiversity(df, group1, group2, metrics=['alpha', 'beta'])
     # Basic assertions
-    assert isinstance(results, pd.DataFrame), "Results should be a DataFrame"
+    assert isinstance(results, tuple(pd.DataFrame, pd.DataFrame)), "Results should be a tuple of two DataFrames"
     assert 'Metric' in results.columns, "Results should have a Metric column"
     assert 'p-val' in results.columns, "Results should have a p-val column"
     # Additional assertions to verify realistic results