move correct_bonferroni into a new module and add correct_holm

examples/correct_multiple_correlations.py

@@ -12,7 +12,8 @@
 import numpy as np
 from scipy.stats import pearsonr, spearmanr
 
-from pypermut.misc import correct_bonferroni, print_results, print_pvals
+from pypermut.correction import correct_bonferroni
+from pypermut.misc import print_results, print_pvals
 from pypermut.stats import permutation_corr
 
 

pypermut/correction.py

@@ -0,0 +1,62 @@
+"""Correction functions."""
+
+import numpy as np
+
+
+def correct_bonferroni(p_vals, n_tests=None):
+    """Correct p-values by Bonferroni's method.
+
+    Correction for multiple tests, using Bonferroni's method [1]_:
+    multiply p-values by the number of tests.
+
+    Parameters
+    ----------
+    p_vals : array, shape (n_vals,)
+        The p-values.
+
+    n_tests : None | int, default=None
+        Number of tests. If None, n_tests is set to n_vals.
+
+    Returns
+    -------
+    p_vals_corrected : array, shape (n_vals,)
+        The p-values corrected by Bonferroni's method.
+
+    References
+    ----------
+    .. [1] https://en.wikipedia.org/wiki/Bonferroni_correction
+    """
+    if n_tests is None:
+        n_tests = len(p_vals)
+    return p_vals * n_tests
+
+
+def correct_holm(p_vals):
+    """Correct p-values by Holm's step-down method.
+
+    Correction for multiple tests, using Holm's step-down method [1]_.
+
+    Parameters
+    ----------
+    p_vals : array, shape (n_vals,)
+        The p-values.
+
+    Returns
+    -------
+    p_vals_corrected : array, shape (n_vals,)
+        The p-values corrected by Holm's method.
+
+    References
+    ----------
+    .. [1] https://en.wikipedia.org/wiki/Holm%E2%80%93Bonferroni_method
+    """
+    n_tests = len(p_vals)
+    sortind = np.argsort(p_vals)
+    p_vals = np.take(p_vals, sortind)
+
+    pvals_corrected_raw = p_vals * np.arange(n_tests, 0, -1)
+    pvals_corrected = np.maximum.accumulate(pvals_corrected_raw)
+    pvals_corrected_ = np.empty_like(pvals_corrected)
+    pvals_corrected_[sortind] = pvals_corrected
+
+    return pvals_corrected_

pypermut/misc.py

@@ -78,34 +78,6 @@ def pvals_to_stars(
     return stars
 
 
-def correct_bonferroni(p_vals, n_tests=None):
-    """Correct p-values by Bonferroni's method.
-
-    Correction for multiple tests, using Bonferroni's method:
-    multiply p-values by the number of tests.
-
-    Parameters
-    ----------
-    p_vals : array, shape (n_vals,)
-        The p-values.
-
-    n_tests : None | int, default=None
-        Number of tests. If None, n_tests is set to n_vals.
-
-    Returns
-    -------
-    p_vals_corrected : array, shape (n_vals,)
-        The p-values corrected by Bonferroni's method.
-
-    References
-    ----------
-    .. [1] https://en.wikipedia.org/wiki/Bonferroni_correction
-    """
-    if n_tests is None:
-        n_tests = len(p_vals)
-    return p_vals * n_tests
-
-
 def print_results(results, r_labels, stat_label):
     """Print results of several tests: statistic value and the p-value.