Compute average_batch_r2 per cell type (#127)

* compute average_batch_r2 per cell type

* Pass the cell type subsets to batch_r2() instead of the donor-level views
* Correct the cell count in the skip comment

* update changelog

* fix changelog spacing

---------

Co-authored-by: Givanna Putri <givanna.h@gmail.com>

Author: rcannood

CHANGELOG.md

@@ -248,3 +248,7 @@
   as scalars instead of lists, as required by `file_score.yaml` (PR #130).
 
 * Clean up stale mock parameters and dead code (PR #133).
+
+
+* Fix bug in `average_batch_r2` where the R2 was computed on all cell types of a donor at once
+  instead of on each cell type separately (PR #127).

src/metrics/average_batch_r2/script.py

@@ -60,13 +60,13 @@
         s2_view_ct = s2_view[s2_view.obs["cell_type"] == ct]
         if (
             s1_view_ct.shape[0] < 20 or s2_view_ct.shape[0] < 20
-        ):  # Skip Rˆ2 calculation if there are less than 10 cells
+        ):  # Skip Rˆ2 calculation if there are less than 20 cells
             print(
                 f"Warning: Rˆ2 not computed for donor {donor} cell type {ct}. Too few cells were present: {s1_view_ct.shape[0]} for split 1 and {s2_view_ct.shape[0]} for split 2"
             )
             continue
 
-        r2_list, marker_list = batch_r2(s1_view, s2_view)
+        r2_list, marker_list = batch_r2(s1_view_ct, s2_view_ct)
 
         marker_list = [ct + "_" + donor + "_" + x for x in marker_list]
         r2_info = [*r2_info, *marker_list]