Update app.py

app.py

@@ -270,6 +270,13 @@ with top_tab_bp:
 
     with sub_cluster:
         st.subheader("Supervised SHAP Clustering (PCA)")
+        st.markdown(
+            "Each point is a gene positioned using PCA on its SHAP profile for the **Most Likely** class, "
+            "so clustering reflects similarity in model explanation patterns rather than raw feature values. "
+            "Genes that appear close together are being prioritised for similar reasons by the model, while genes far apart "
+            "are driven by different feature contributions. Use the overlays to see whether your queried genes align with "
+            "the most-likely training-gene region or form distinct groups."
+        )
 
         try:
             training_genes = pd.read_csv("training_cleaned.csv")