app: warn if input values are out of log2 range (guards against raw TPM/counts)

app.py

@@ -96,6 +96,21 @@ if uploaded:
             if overlap < 1000:
                 st.warning("Low gene overlap — results may be unreliable.")
 
+            # --- Input scale check: model expects log2(TPM) ---
+            common_for_scale = [g for g in gene_list if g in df.columns]
+            if common_for_scale:
+                vals = df[common_for_scale].to_numpy(dtype=float)
+                vmax = float(np.nanmax(vals))
+                p99 = float(np.nanpercentile(vals, 99))
+                st.caption(f"Input value range: 99th pct = {p99:.1f}, max = {vmax:.1f} "
+                           "(log2(TPM) is typically 0–20).")
+                if vmax > 30 or p99 > 25:
+                    st.warning(
+                        "⚠️ Input values look larger than expected for **log2 scale**. "
+                        "This model expects **log2-transformed expression** (e.g. log2(TPM+1)). "
+                        "Raw TPM/counts will give unreliable predictions — please log2-transform first."
+                    )
+
             aligned = pd.DataFrame(0.0, index=df.index, columns=gene_list)
             common = [g for g in gene_list if g in df.columns]
             aligned[common] = df[common].values