Add caching to the instance

app.py

@@ -1,4 +1,4 @@
-import uuid
+import hashlib
 import gradio as gr
 from pathlib import Path
 from Bio.PDB import MMCIFParser, PDBIO
@@ -51,20 +51,25 @@ def predict(sequence: str) -> str:
     seq_file = Path("sequence.fasta")
     _write_fasta_file(seq_file, sequence)
     
-    # Set up unique output directory
-    seq_id = str(uuid.uuid4())
+    # Set up unique output directory based on sequence hash
+    seq_id = hashlib.sha1(sequence.encode()).hexdigest()
     output_dir = Path(f"sequence_{seq_id}")
     output_dir.mkdir(parents=True, exist_ok=True)
     
-    # Run Boltz prediction
-    logger.info(f"Predicting {seq_file.stem} with project code {FOLDING_PROJECT_CODE}")
-    boltz_predict(
-        source=seq_file,
-        project_code=FOLDING_PROJECT_CODE, 
-        output=output_dir,
-        unzip=True
-    )
-    logger.info("Prediction done. Output directory: %s", output_dir)
+    # Check if prediction already exists
+    pred_cif = list(output_dir.rglob("*_model_0.cif"))
+    if not pred_cif:
+        # Run Boltz prediction
+        logger.info(f"Predicting {seq_file.stem} with project code {FOLDING_PROJECT_CODE}")
+        boltz_predict(
+            source=seq_file,
+            project_code=FOLDING_PROJECT_CODE, 
+            output=output_dir,
+            unzip=True
+        )
+        logger.info("Prediction done. Output directory: %s", output_dir)
+    else:
+        logger.info("Prediction already exists. Output directory: %s", output_dir)
 
     # # TODO: remove this
     # output_dir = Path("boltz_results")