fix bug when max_n_pl<n_pl

Author: bquistorff

src/SparseSC/utils/metrics_utils.py

@@ -290,7 +290,7 @@ def _ncr(n, r):
         # math a bit nicer, I will reject a hypothesis if pval<=(1-level)
         assert 0 < level < 1 and level > 0, "Use a level in [0,1]"
         alpha = 1 - level
-        p2min = 2 / n_pl
+        p2min = 2 / comb_len
         alpha_ind = max((1, round(alpha / p2min))) - 1
         alpha = alpha_ind * p2min