wip: scatter plots

app.py

@@ -192,6 +192,7 @@ def fake_predict_and_correlate(spr_data_with_scores: pd.DataFrame, score_cols: l
     """Fake predict structures of all complexes and correlate the results."""
     corr_data = []
     for score_col in score_cols:
+        logging.info("Computing correlation between %s and KD (nM)", score_col)
         res = spearmanr(spr_data_with_scores["KD (nM)"], spr_data_with_scores[score_col])
         corr_data.append({"score": score_col, "correlation": res.statistic, "p-value": res.pvalue})
 
@@ -200,7 +201,7 @@ def fake_predict_and_correlate(spr_data_with_scores: pd.DataFrame, score_cols: l
     corr_data = corr_data.sort_values('correlation', ascending=True)
     
     # Create bar plot of correlations
-    corr_plot = go.Figure(data=[
+    corr_ranking_plot = go.Figure(data=[
         go.Bar(
             x=corr_data["correlation"],
             y=corr_data["score"], 
@@ -209,7 +210,7 @@ def fake_predict_and_correlate(spr_data_with_scores: pd.DataFrame, score_cols: l
             hovertemplate="<i>Score:</i> %{y}<br><i>Correlation:</i> %{x:.3f}<br>"
         )
     ])
-    corr_plot.update_layout(
+    corr_ranking_plot.update_layout(
         title="Correlation with Binding Affinity",
         yaxis_title="Score Type",
         xaxis_title="Spearman Correlation",
@@ -217,9 +218,24 @@ def fake_predict_and_correlate(spr_data_with_scores: pd.DataFrame, score_cols: l
         showlegend=False
     )
 
+    # corr_plot is a scatter plot of the correlation between the binding affinity and each of the scores
+    scatters = []
+    for score_col in score_cols:
+        scatters.append(
+            go.Scatter(
+                x=spr_data_with_scores["KD (nM)"],
+                y=spr_data_with_scores[score_col],
+                name=f"KD (nM) vs {score_col}",
+                mode='markers',  # Only show markers/dots, no lines
+                hovertemplate="<i>Score:</i> %{y}<br><i>KD (nM):</i> %{x:.2f}<br>"
+            )   
+        )
+    corr_plot = go.Figure(data=scatters)
+    
     cols_to_show = ["KD (nM)"]
     cols_to_show.extend(score_cols)
-    return spr_data_with_scores[cols_to_show], corr_plot
+
+    return spr_data_with_scores[cols_to_show], corr_ranking_plot, corr_plot
 
 demo = gr.Blocks(title="Folding Studio: structure prediction with Boltz-1")
 
@@ -261,7 +277,9 @@ with demo:
                     with gr.Row():
                         prediction_dataframe = gr.Dataframe(label="Predicted Structures Data")
                 with gr.Column():
-                    correlation_plot = gr.Plot(label="Correlation")
+                    correlation_ranking_plot = gr.Plot(label="Correlation ranking")
+                    correlation_plot = gr.Plot(label="Correlation with binding affinity")
+
             
 
             cols = [
@@ -297,7 +315,7 @@ with demo:
             fake_predict_btn.click(
                 fn=lambda x: fake_predict_and_correlate(spr_data_with_scores, cols),
                 inputs=None,
-                outputs=[prediction_dataframe, correlation_plot]
+                outputs=[prediction_dataframe, correlation_ranking_plot, correlation_plot]
             )
 
     

pyproject.toml

@@ -7,6 +7,8 @@ requires-python = ">=3.11"
 dependencies = [
     "gradio==5.30.0",
     "ipython>=9.2.0",
+    "jupyter>=1.1.1",
     "numpy>=2.2.6",
     "plotly>=6.1.1",
+    "scipy>=1.15.3",
 ]