Update app.py

app.py

@@ -202,71 +202,84 @@ with block:
         # 1️⃣  Leaderboard tab
         # ------------------------------------------------------------------
         with gr.TabItem("🏅 PROBE Leaderboard", elem_id="probe-benchmark-tab-table", id=1):
-            # small workflow figure at top
+        
+            # ── header ────────────────────────────────────────────────────
             gr.Image(
                 value="./src/data/PROBE_workflow_figure.jpg",
                 show_label=False,
                 height=1000,
                 container=False,
             )
-
             gr.Markdown(
                 "## For detailed explanations of the metrics and benchmarks, please refer to the 📝 About tab.",
                 elem_classes="leaderboard-note",
             )
-
-            leaderboard = get_baseline_df(None, None)
-            method_names = leaderboard['Method'].unique().tolist()
-            metric_names = leaderboard.columns.tolist(); metric_names.remove('Method')
-
+        
+            # ── data prep ────────────────────────────────────────────────
+            leaderboard      = get_baseline_df(None, None)
+            method_names     = leaderboard["Method"].unique().tolist()
+            metric_names     = leaderboard.columns.tolist(); metric_names.remove("Method")
+        
+            base_method_names = [m for m in method_names if m in base_methods]
+            user_method_names = [m for m in method_names if m not in base_methods]
+        
             benchmark_metric_mapping = {
-                "Semantic Similarity Inference": [m for m in metric_names if m.startswith('sim_')],
-                "Ontology-based Protein Function Prediction":  [m for m in metric_names if m.startswith('func')],
-                "Drug Target Protein Family Classification":    [m for m in metric_names if m.startswith('fam_')],
-                "Protein-Protein Binding Affinity Estimation":  [m for m in metric_names if m.startswith('aff_')],
+                "Semantic Similarity Inference":               [m for m in metric_names if m.startswith("sim_")],
+                "Ontology-based Protein Function Prediction":  [m for m in metric_names if m.startswith("func")],
+                "Drug Target Protein Family Classification":   [m for m in metric_names if m.startswith("fam_")],
+                "Protein-Protein Binding Affinity Estimation": [m for m in metric_names if m.startswith("aff_")],
             }
-
-            leaderboard_method_selector = gr.CheckboxGroup(
-                choices=method_names,
-                label="Select Methods",
-                value=method_names,
-                interactive=True,
-            )
-
+        
+            # ── callback helper ──────────────────────────────────────────
+            def update_leaderboard_combined(selected_base, selected_user, selected_metrics):
+                selected_methods = (selected_base or []) + (selected_user or [])
+                return build_leaderboard_styler(selected_methods, selected_metrics)
+        
+            # ── collapsible selectors ────────────────────────────────────
+            with gr.Accordion("📦 Base methods", open=False):
+                leaderboard_method_selector_base = gr.CheckboxGroup(
+                    choices=base_method_names,
+                    label="Base Methods",
+                    value=base_method_names,          # ← all selected
+                    interactive=True,
+                )
+        
+            with gr.Accordion("🛠️ User-defined methods", open=False):
+                leaderboard_method_selector_user = gr.CheckboxGroup(
+                    choices=user_method_names,
+                    label="User Methods",
+                    value=[],                         # ← none selected
+                    interactive=True,
+                )
+        
             benchmark_type_selector_lb = gr.CheckboxGroup(
                 choices=list(benchmark_metric_mapping.keys()),
-                label="Select Benchmark Types",
-                value=None,
+                label="Benchmark Types",
+                value=list(benchmark_metric_mapping.keys()),  # ← all selected
                 interactive=True,
             )
-
-            leaderboard_metric_selector = gr.CheckboxGroup(
-                choices=metric_names,
-                label="Select Metrics",
-                value=None,
-                interactive=True,
-            )
-
-            baseline_value = get_baseline_df(method_names, metric_names)
-            baseline_value = baseline_value.applymap(lambda x: round(x, 4) if isinstance(x, (int, float)) else x)
-            baseline_header  = ["Method"] + metric_names
-            baseline_datatype = ['markdown'] + ['number'] * len(metric_names)            
-
-            styler = build_leaderboard_styler()
-
+        
+            with gr.Accordion("📐 Metrics", open=False):
+                leaderboard_metric_selector = gr.CheckboxGroup(
+                    choices=metric_names,
+                    label="Select Metrics",
+                    value=metric_names,               # ← all selected
+                    interactive=True,
+                )
+        
+            # ── colour / shading legend (unchanged) ──────────────────────
             with gr.Row():
                 with gr.Column(scale=1):
                     gr.Markdown(
                         """
                     ## Method-name colours  
                     <span style='color:green; font-weight:bold; font-size:1.1rem;'>🟢  Classical representations  
-                    <span style='color:blue;  font-weight:bold; font-size:1.1rem;'>🔵  Small-scale Protein Language Models  
-                    <span style='color:red;   font-weight:bold; font-size:1.1rem;'>🔴  Large-scale Protein Language Models  
-                    <span style='color:orange;font-weight:bold; font-size:1.1rem;'>🟠  Multimodal Protein Language Models
+                    <span style='color:blue;  font-weight:bold; font-size:1.1rem;'>🔵  Small-scale Protein LMs  
+                    <span style='color:red;   font-weight:bold; font-size:1.1rem;'>🔴  Large-scale Protein LMs  
+                    <span style='color:orange;font-weight:bold; font-size:1.1rem;'>🟠  Multimodal Protein LMs
                     """,
                         elem_classes="leaderboard-note",
                     )
-
                 with gr.Column(scale=1):
                     gr.Markdown(
                         """
@@ -287,23 +300,36 @@ with block:
                         elem_classes="leaderboard-note",
                     )
 
-            with gr.Row(show_progress=True, variant='panel'):
-                data_component = gr.Dataframe(
-                    value=styler,
-                    headers=baseline_header,
-                    type="pandas",
-                    datatype=baseline_datatype,
-                    interactive=False,
-                    elem_id="leaderboard-table",
-                    pinned_columns=1,
-                    max_height=1000,
-                    show_fullscreen_button=True
-                )
-
-            # callbacks
-            leaderboard_method_selector.change(
-                update_leaderboard,
-                inputs=[leaderboard_method_selector, leaderboard_metric_selector],
+            gr.Markdown("### If a method name ends with (`^`), it suggests potential suspicions of data leakage related to **similarity**, **function**, or **family** benchmarks.")
+            # ── dataframe ────────────────────────────────────────────────
+            styler = build_leaderboard_styler(base_method_names, metric_names)
+        
+            data_component = gr.Dataframe(
+                value=styler,
+                headers=["Method"] + metric_names,
+                type="pandas",
+                datatype=["markdown"] + ["number"] * len(metric_names),
+                interactive=False,
+                elem_id="leaderboard-table",
+                pinned_columns=1,
+                max_height=1000,
+                show_fullscreen_button=True,
+            )
+        
+            # ── callbacks ────────────────────────────────────────────────
+            leaderboard_method_selector_base.change(
+                update_leaderboard_combined,
+                inputs=[leaderboard_method_selector_base, leaderboard_method_selector_user, leaderboard_metric_selector],
+                outputs=data_component,
+            )
+            leaderboard_method_selector_user.change(
+                update_leaderboard_combined,
+                inputs=[leaderboard_method_selector_base, leaderboard_method_selector_user, leaderboard_metric_selector],
+                outputs=data_component,
+            )
+            leaderboard_metric_selector.change(
+                update_leaderboard_combined,
+                inputs=[leaderboard_method_selector_base, leaderboard_method_selector_user, leaderboard_metric_selector],
                 outputs=data_component,
             )
             benchmark_type_selector_lb.change(
@@ -311,12 +337,6 @@ with block:
                 inputs=[benchmark_type_selector_lb],
                 outputs=leaderboard_metric_selector,
             )
-            leaderboard_metric_selector.change(
-                update_leaderboard,
-                inputs=[leaderboard_method_selector, leaderboard_metric_selector],
-                outputs=data_component,
-            )
-
         # ------------------------------------------------------------------
         # 2��⃣  Visualisation tab
         # ------------------------------------------------------------------