Changed normalization to weights (1=n)

scverse · Mar 28, 2020 · 0875529 · 0875529
1 parent db501c8
commit 0875529
Show file tree

Hide file tree

Showing 3 changed files with 4 additions and 4 deletions.
diff --git a/scirpy/_tools/_group_abundance.py b/scirpy/_tools/_group_abundance.py
@@ -18,7 +18,7 @@ def _group_abundance(
 
     # normalize to fractions
     scale_vector = _normalize_counts(tcr_obs, normalize=fraction, default_col=groupby)
-    tcr_obs = tcr_obs.assign(count=1, weight=1 / scale_vector)
+    tcr_obs = tcr_obs.assign(count=1, weight=scale_vector)
 
     # Calculate distribution of lengths in each group. Use sum instead of count
     # to reflect weights

diff --git a/scirpy/_tools/_vdj_usage.py b/scirpy/_tools/_vdj_usage.py
@@ -44,10 +44,10 @@ def vdj_usage(
     """
 
     if fraction is None:
-        fraction = 1 / _normalize_counts(adata.obs, normalize=False)
+        fraction = _normalize_counts(adata.obs, normalize=False)
     else:
         if isinstance(fraction, (bool, str)):
-            fraction = 1 / _normalize_counts(adata.obs, normalize=fraction)
+            fraction = _normalize_counts(adata.obs, normalize=fraction)
 
     observations = adata.obs.loc[:, target_cols]
     observations[size_column] = fraction

diff --git a/scirpy/_util/__init__.py b/scirpy/_util/__init__.py
@@ -160,7 +160,7 @@ def _normalize_counts(
         raise ValueError("No colname specified in either `normalize` or `default_col")
 
     # https://stackoverflow.com/questions/29791785/python-pandas-add-a-column-to-my-dataframe-that-counts-a-variable
-    return obs.groupby(normalize_col)[normalize_col].transform("count").values
+    return 1 / obs.groupby(normalize_col)[normalize_col].transform("count").values
 
 
 def _get_from_uns(adata: AnnData, tool: str, *, parameters: dict = None) -> Any: