Added new metrics (here and to dataset), and loaded results once

app.py

@@ -57,7 +57,7 @@ def make_submission(
     return "✅ Your submission has been received! Sit tight and your scores will appear on the leaderboard shortly."
 
     
-def get_leaderboard_table(assay: str | None = None):
+def get_leaderboard_table(df_results: pd.DataFrame, assay: str | None = None):
     # ds = load_dataset(results_repo, split='train', download_mode="force_redownload")
     # full_df = pd.DataFrame(ds)
     # full_df['full results'] = full_df['result_filename'].apply(lambda x: make_boundary_clickable(x)).astype(str)
@@ -69,18 +69,15 @@ def get_leaderboard_table(assay: str | None = None):
     # to_show['user'] = to_show['user'].apply(lambda x: make_user_clickable(x)).astype(str)
     
     # Previously hosted on HF hub, local for now (Can also pull directly from github backend)
-    column_order = ["model", "property", "spearman", "spearman_abs"]  # "assay", 
-    ds = load_dataset(results_repo, split='no_low_spearman', download_mode="force_redownload")
-    df = pd.DataFrame(ds).drop_duplicates(subset=["model", "assay"])
-    df["property"] = df["assay"].map(ASSAY_RENAME)
-    df = df.query("assay.isin(@ASSAY_RENAME.keys())")
+    column_order = ["model", "property", "spearman", "spearman_abs"]
+    df = df_results.query("assay.isin(@ASSAY_RENAME.keys())").copy()
     if assay is not None:
         df = df[df['assay'] == assay]
     df = df[column_order]
     return df.sort_values(by="spearman_abs", ascending=False)
 
-def get_leaderboard_object(assay: str | None = None):
-    df = get_leaderboard_table(assay=assay)
+def get_leaderboard_object(df_results: pd.DataFrame, assay: str | None = None):
+    df = get_leaderboard_table(df_results=df_results, assay=assay)
     filter_columns = ["model"]
     if assay is None:
         filter_columns.append("property")
@@ -98,28 +95,31 @@ def get_leaderboard_object(assay: str | None = None):
 def show_output_box(message):
     return gr.update(value=message, visible=True)
 
-# 
-# def gradio_interface() -> gr.Blocks:
+def fetch_hf_results():
+    ds = load_dataset(results_repo, split='no_low_spearman', download_mode="force_redownload")
+    df = pd.DataFrame(ds).drop_duplicates(subset=["model", "assay"])
+    df["property"] = df["assay"].map(ASSAY_RENAME)
+    return df
+
 with gr.Blocks() as demo:
     gr.Markdown("""
         ## Welcome to the Ginkgo Antibody Developability Benchmark Leaderboard!
         
         Participants can submit their model to the leaderboard by 
         """)
+    df = fetch_hf_results()
     with gr.Tabs(elem_classes="tab-buttons"):
-        with gr.TabItem("🚀 Leaderboard", elem_id="abdev-benchmark-tab-table"):                
-            gr.Markdown("# Antibody Developability Benchmark Leaderboard")
-
-            get_leaderboard_object()
-            # TODO: this is not going to update well, need to fix
-
-            # gr.Markdown("Extra info here")
-
         # Procedurally make these 5 tabs
         for assay in ASSAY_LIST:
             with gr.TabItem(f"{ASSAY_EMOJIS[assay]} {ASSAY_RENAME[assay]}", elem_id=f"abdev-benchmark-tab-table"):
                 gr.Markdown(f"# {ASSAY_RENAME[assay]} (measured by {assay})")
-                get_leaderboard_object(assay=assay)
+                get_leaderboard_object(df_results=df, assay=assay)
+        
+        with gr.TabItem("🚀 Overall", elem_id="abdev-benchmark-tab-table"):                
+            gr.Markdown("# Antibody Developability Benchmark Leaderboard over all properties")
+
+            get_leaderboard_object(df_results=df)
+            # TODO: this is not going to update well, need to fix
         
         with gr.TabItem("❔About", elem_id="abdev-benchmark-tab-table"):
             gr.Markdown(

data/metrics_all.csv

@@ -1,21 +1,26 @@
-spearman,assay,model,spearman_abs
-0.422834774225429,HIC,Aggrescan3D - aggrescan_average_score,0.422834774225429
-0.358762795727933,AC-SINS_pH7.4,TAP - PNC,0.358762795727933
-0.3585224061081473,HIC,Aggrescan3D - aggrescan_90_score,0.3585224061081473
-0.3365516014806938,PR_CHO,Saprot_VH - solubility_probability,0.3365516014806938
-0.3203773185543964,AC-SINS_pH7.4,TAP - SFvCSP,0.3203773185543964
-0.3044160918625593,HIC,Aggrescan3D - aggrescan_max_score,0.3044160918625593
-0.2450651623577951,HIC,TAP - SFvCSP,0.2450651623577951
-0.2381972244142228,PR_CHO,TAP - SFvCSP,0.2381972244142228
-0.1924791603648384,Tm2,Saprot_VH - stability_score,0.1924791603648384
-0.1923458958277369,HIC,TAP - CDR Length,0.1923458958277369
-0.1878766623808878,Titer,AntiFold,0.1878766623808878
-0.18059398754127,HIC,DeepViscosity,0.18059398754127
-0.1691412287169806,AC-SINS_pH7.4,TAP - PPC,0.1691412287169806
-0.151234196032203,PR_CHO,TAP - PNC,0.151234196032203
-0.1501689804134715,AC-SINS_pH7.4,TAP - CDR Length,0.1501689804134715
-0.1423756688786398,Titer,TAP - PPC,0.1423756688786398
-0.1406309504998865,PR_CHO,Aggrescan3D - aggrescan_max_score,0.1406309504998865
-0.1218057192943458,Tm2,AntiFold,0.1218057192943458
-0.1144051722170351,HIC,hic_model_name,0.1144051722170351
-0.0747719620306879,HIC,Aggrescan3D - aggrescan_cdrh3_average_score,0.0747719620306879
+assay,model,spearman,spearman_cross_val,top_10_recall,top_10_recall_cross_val
+HIC,Aggrescan3D - aggrescan_average_score,0.422834774225429,,0.3333333333333333,
+AC-SINS_pH7.4,TAP - linear regression,0.4019194824087021,0.3401456689218918,0.375,0.2799999999999999
+HIC,TAP - linear regression,0.3622075317102941,0.2222991438172065,0.4166666666666667,0.43
+AC-SINS_pH7.4,TAP - PNC,0.358762795727933,,0.2916666666666667,
+HIC,Aggrescan3D - aggrescan_90_score,0.3585224061081473,,0.2083333333333333,
+PR_CHO,Saprot_VH - solubility_probability,0.3365516014806938,,0.0833333333333333,
+AC-SINS_pH7.4,TAP - SFvCSP,0.3203773185543964,,0.2083333333333333,
+HIC,Aggrescan3D - aggrescan_max_score,0.3044160918625593,,0.2083333333333333,
+PR_CHO,TAP - linear regression,0.260631929274264,0.1560705020744792,0.8333333333333334,0.47
+HIC,TAP - SFvCSP,0.2450651623577951,,0.2083333333333333,
+PR_CHO,TAP - SFvCSP,0.2381972244142228,,0.0,
+Tm2,Saprot_VH - stability_score,0.1924791603648384,,0.1666666666666666,
+HIC,TAP - CDR Length,0.1923458958277369,,0.0833333333333333,
+Titer,AntiFold,0.1878766623808878,,0.0833333333333333,
+HIC,DeepViscosity,0.18059398754127,,0.0416666666666666,
+AC-SINS_pH7.4,TAP - PPC,0.1691412287169806,,0.0833333333333333,
+Titer,TAP - linear regression,0.1682403605307924,0.1129210260701206,0.3333333333333333,0.38
+PR_CHO,TAP - PNC,0.151234196032203,,0.0416666666666666,
+AC-SINS_pH7.4,TAP - CDR Length,0.1501689804134715,,0.0,
+Titer,TAP - PPC,0.1423756688786398,,0.0833333333333333,
+PR_CHO,Aggrescan3D - aggrescan_max_score,0.1406309504998865,,0.0833333333333333,
+Tm2,AntiFold,0.1218057192943458,,0.125,
+HIC,hic_model_name,0.1144051722170351,0.1511895582680471,0.1666666666666666,0.05
+Tm2,TAP - linear regression,0.0844935706523633,-0.1153965363405958,0.6666666666666666,0.64
+HIC,Aggrescan3D - aggrescan_cdrh3_average_score,0.0747719620306879,,0.25,