Added not on fully overlapping marker genes between cluster pairs.

tutorials/cytocipher_pancreas.ipynb

@@ -196,6 +196,7 @@
    ]
   },
   {
+   "attachments": {},
    "cell_type": "markdown",
    "id": "908de77a",
    "metadata": {},
@@ -206,11 +207,17 @@
     "\n",
     "#### NOTE on determining if your data is over-clustered\n",
     "\n",
-    "    * Practically, Leiden clustering at a resolution between 2 and 4 produces over-clusters.\n",
-    "    * UMAP and the Cytocipher enrichment heatmap can be utilised to assess over-clustering. \n",
-    "    * In the UMAP, over-clusters will display as subclusters of larger populations with no apparent separation. \n",
-    "    * In the enrichment heatmap, cross-scoring is indicated by scores along each column split amongst multiple rows (indicating cells in separate clusters having similar scores). \n",
-    "    * The below heatmap is a good example, though still has some cross-scoring after significance analysis due to the continuous nature of developmental scRNA-seq data."
+    "    • Practically, Leiden clustering at a resolution between 2 and 4 produces over-clusters.\n",
+    "    • UMAP and the Cytocipher enrichment heatmap can be utilised to assess over-clustering. \n",
+    "    • In the UMAP, over-clusters will display as subclusters of larger populations with no apparent separation. \n",
+    "    • In the enrichment heatmap, cross-scoring is indicated by scores along each column split amongst multiple rows (indicating cells in separate clusters having similar scores). \n",
+    "    • The below heatmap is a good example, though still has some cross-scoring after significance analysis due to the continuous nature of developmental scRNA-seq data.\n",
+    "    \n",
+    "#### NOTE on cluster pairs with fully overlapping markers genes\n",
+    "\n",
+    "    • Full overlap of marker genes between a pair of clusters (e.g. cluster i and cluster j) results in no set difference between them, and therefore negative gene set subtraction with respect to cluster j is not performed for cluster i. \n",
+    "    • Clusters with full overlap are scored/merged/kept separate the same as clusters with no overlaps (albeit scored for the same set of genes in the bi-directional test). \n",
+    "    • Optionally, the user can run with squash_exception=False when running cc.tl.merge_clusters or cc.tl.code_score, to receive an error message if full gene set overlap occurs between two cluster pairs. This prompts the user to increase the number of marker genes per cluster, in-case additional genes discriminate the cluster pair."
    ]
   },
   {
@@ -844,7 +851,7 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "Python 3",
+   "display_name": "Python 3 (ipykernel)",
    "language": "python",
    "name": "python3"
   },
@@ -858,7 +865,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.12"
+   "version": "3.8.16"
   }
  },
  "nbformat": 4,