debase 0.1.17__tar.gz → 0.1.19__tar.gz

{debase-0.1.17 → debase-0.1.19}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: debase
-Version: 0.1.17
+Version: 0.1.19
 Summary: Enzyme lineage analysis and sequence extraction package
 Home-page: https://github.com/YuemingLong/DEBase
 Author: DEBase Team

{debase-0.1.17 → debase-0.1.19}/src/debase/_version.py

@@ -1,3 +1,3 @@
 """Version information."""
 
-__version__ = "0.1.17"
+__version__ = "0.1.19"

{debase-0.1.17 → debase-0.1.19}/src/debase/reaction_info_extractor.py

@@ -927,12 +927,77 @@ Ignore locations that contain data for other campaigns.
     # ------------------------------------------------------------------
     # 6.3 Extract metrics in batch
     # ------------------------------------------------------------------
+    
+    def _validate_location_exists(self, ref: str) -> bool:
+        """Verify that the referenced location actually exists in the document."""
+        # Search for the actual reference in the document
+        for page_num in range(len(self.doc)):
+            page = self.doc[page_num]
+            text = page.get_text()
+            
+            # Look for table references like "Table 1", "Table S1", etc.
+            if re.search(rf'\b{re.escape(ref)}\b', text, re.IGNORECASE):
+                return True
+        
+        return False
+
+    def _validate_context(self, snippet: str, enzyme_list: List[str], ref: str) -> bool:
+        """Validate that the context contains meaningful content for extraction."""
+        if not snippet or len(snippet.strip()) < 50:
+            LOGGER.warning("Insufficient context for extraction from %s - skipping", ref)
+            return False
+        
+        # Check if context actually mentions the enzymes we're looking for
+        enzyme_mentions = sum(1 for enzyme in enzyme_list if enzyme.lower() in snippet.lower())
+        if enzyme_mentions == 0:
+            LOGGER.warning("No enzyme mentions found in context for %s - skipping", ref)
+            return False
+        
+        # Check for performance-related keywords
+        performance_keywords = ['yield', 'selectivity', 'conversion', 'ee', 'er', 'ttn', 'ton', 'tof', '%', 'percent']
+        has_performance_data = any(keyword in snippet.lower() for keyword in performance_keywords)
+        
+        if not has_performance_data:
+            LOGGER.warning("No performance metrics found in context for %s - skipping", ref)
+            return False
+        
+        LOGGER.info("Context validated for %s: %d chars, %d enzyme mentions", ref, len(snippet), enzyme_mentions)
+        return True
+
+    def _validate_response(self, data: Dict, enzyme_list: List[str], ref: str) -> bool:
+        """Validate that the response contains meaningful data for the requested enzymes."""
+        if not data or not isinstance(data, dict):
+            LOGGER.warning("Invalid response format from %s - skipping", ref)
+            return False
+        
+        # Check if we got data for at least one enzyme
+        enzymes_with_data = 0
+        for enzyme in enzyme_list:
+            enzyme_data = data.get(enzyme, {})
+            if isinstance(enzyme_data, dict) and enzyme_data:
+                # Check if there's at least one non-null metric
+                metrics = ['yield', 'ttn', 'ton', 'selectivity', 'conversion', 'tof', 'activity']
+                has_metric = any(enzyme_data.get(metric) is not None for metric in metrics)
+                if has_metric:
+                    enzymes_with_data += 1
+        
+        if enzymes_with_data == 0:
+            LOGGER.warning("No valid metrics found in response from %s - skipping", ref)
+            return False
+        
+        LOGGER.info("Response validated for %s: %d enzymes with data", ref, enzymes_with_data)
+        return True
 
     def extract_metrics_batch(self, enzyme_list: List[str], ref: str) -> List[Dict[str, Any]]:
         """Extract performance metrics for multiple enzymes from the identified location in batch."""
         ref_lc = ref.lower()
         image_b64: Optional[str] = None
         
+        # First, validate that the location actually exists in the document
+        if not self._validate_location_exists(ref):
+            LOGGER.warning("Location %s not found in document - skipping", ref)
+            return []
+        
         # Add campaign context if available
         campaign_context = ""
         if self.campaign_filter:
@@ -953,6 +1018,10 @@ Ignore locations that contain data for other campaigns.
         else:
             snippet = self._page_with_reference(ref) or ""
 
+        # Validate context before sending to Gemini
+        if not image_b64 and not self._validate_context(snippet, enzyme_list, ref):
+            return []
+
         enzyme_names = "\n".join([f"- {enzyme}" for enzyme in enzyme_list])
         
         if image_b64:
@@ -961,8 +1030,9 @@ Ignore locations that contain data for other campaigns.
             LOGGER.info("Gemini Vision: extracting metrics for %d enzymes from %s…", len(enzyme_list), ref)
             tag = f"extract_metrics_batch_vision"
         else:
-            # Add enzyme names to prompt for batch extraction
-            prompt = campaign_context + PROMPT_EXTRACT_METRICS + f"\n\nExtract performance data for ALL these enzyme variants:\n{enzyme_names}\n\n=== CONTEXT ===\n" + snippet[:4000]
+            # Add enzyme names to prompt for batch extraction with explicit format requirement
+            format_example = '{"enzyme1": {"yield": "99.0%", "ttn": null, ...}, "enzyme2": {"yield": "85.0%", ...}}'
+            prompt = campaign_context + PROMPT_EXTRACT_METRICS + f"\n\nExtract performance data for ALL these enzyme variants:\n{enzyme_names}\n\nReturn a JSON object with enzyme names as keys, each containing the metrics.\nExample format: {format_example}\n\n=== CONTEXT ===\n" + snippet[:4000]
             LOGGER.info("Gemini: extracting metrics for %d enzymes from %s…", len(enzyme_list), ref)
             tag = f"extract_metrics_batch"
 
@@ -976,6 +1046,10 @@ Ignore locations that contain data for other campaigns.
                 image_b64=image_b64
             )
             
+            # Validate response has meaningful data
+            if not self._validate_response(data, enzyme_list, ref):
+                return []
+            
             # Handle the response format - expecting a dict with enzyme names as keys
             results = []
             if isinstance(data, dict):

{debase-0.1.17 → debase-0.1.19}/src/debase.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: debase
-Version: 0.1.17
+Version: 0.1.19
 Summary: Enzyme lineage analysis and sequence extraction package
 Home-page: https://github.com/YuemingLong/DEBase
 Author: DEBase Team