tune down cursor

folding_studio_demo/app.py

@@ -275,26 +275,27 @@ def model_comparison(api_key: str) -> None:
 
 
 def create_antibody_discovery_tab():
-    gr.Markdown("# Accelerating Antibody Discovery: In-Silico and Experimental Insights")
+    gr.Markdown(
+        "# Accelerating Antibody Discovery: In-Silico and Experimental Insights"
+    )
     gr.Markdown("""
-        Hey there! 👋 Let's dive into how we're using AI to accelerate antibody drug discovery by looking at how protein folding models stack up against real lab data.
+        Let's dive into how we're using AI to accelerate antibody drug discovery by looking at how protein folding models stack up against real lab data.
 
-        We've got this fascinating dataset that shows how well different antibodies stick to a specific target (we measure this as KD in nM). 🧪
+        We've got this dataset that shows how well different antibodies stick to a specific target (we measure this as KD in nM).
         For each antibody-target pair, we've recorded:
-        - The antibody's light and heavy chain sequences (think of them as the antibody's building blocks) 🧬
-        - The target (antigen) sequence 🎯
-        - How strongly they bind together in the lab (the KD value, lower means stronger binding) 💪
+        - The antibody's light and heavy chain sequences (think of them as the antibody's building blocks)
+        - The target (antigen) sequence
+        - How strongly they bind together in the lab (the KD value, lower means stronger binding)
         
-        Here's where it gets interesting! We take these sequences and feed them into protein folding models
+        Why is it interesting? We take these sequences and feed them into protein folding models
         that predict their 3D structures. The models tell us how confident they are about their predictions.
         By comparing these confidence scores with our lab results, we can figure out which model scores
-        are actually good at predicting real binding strength! 🎯
+        are actually good at predicting real binding strength!
         
-        Why is this exciting for drug discovery? 🚀 Once we know which computational scores to trust,
+        Why is this useful for drug discovery? Once we know which computational scores to trust,
         we can use them to quickly check thousands of potential antibodies without having to test each one
-        in the lab. It's like having a high-speed screening tool! We can then focus our lab work on testing
-        just the most promising candidates. This means we can find effective antibody drugs much faster than
-        before! 🔬✨
+        in the lab. We can then focus our lab work on testing just the most promising candidates. 
+        This means we can find effective antibody drugs much faster than before!
     """)
     spr_data_with_scores = pd.read_csv("spr_af_scores_mapped.csv")
     spr_data_with_scores = spr_data_with_scores.rename(columns=SCORE_COLUMN_NAMES)
@@ -323,7 +324,9 @@ def create_antibody_discovery_tab():
 
     with gr.Row():
         with gr.Column(min_width=150):
-            gr.Markdown("Now, let's see how well the protein folding models can predict the binding affinity of these antibodies to the target antigen.")
+            gr.Markdown(
+                "Now, let's see how well the protein folding models can predict the binding affinity of these antibodies to the target antigen."
+            )
         with gr.Column(min_width=150):
             fake_predict_btn = gr.Button(
                 "Predict structures of all complexes",