CASSIA 0.3.1.dev5__tar.gz → 0.3.3__tar.gz

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/__init__.py

@@ -1,7 +1,7 @@
 # CASSIA - Cell Annotation with Semantic Similarity for Intelligent Analysis
 # Root module with backward-compatible exports from reorganized submodules
 
-__version__ = "0.3.1.dev5"
+__version__ = "0.3.3"
 
 # =============================================================================
 # BACKWARD COMPATIBILITY LAYER
@@ -126,6 +126,11 @@ try:
 except ImportError:
     pass
 
+try:
+    from .reports.generate_report_uncertainty import generate_uq_html_report
+except ImportError:
+    pass
+
 # -----------------------------------------------------------------------------
 # PIPELINE
 # -----------------------------------------------------------------------------

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/agents/annotation_boost/annotation_boost.py

@@ -671,6 +671,93 @@ def get_marker_info(gene_list: List[str], marker: Union[pd.DataFrame, Any]) -> s
 
     return marker_string
 
+def extract_gene_stats_from_conversation(conversation_history: List[Dict[str, str]]) -> Dict[str, Dict[str, str]]:
+    """
+    Extract gene statistics from conversation history messages.
+
+    The conversation history contains gene expression data in tabular format from USER messages.
+    This function parses those tables and creates a lookup dictionary for tooltip display.
+
+    Args:
+        conversation_history: List of conversation messages with role and content
+
+    Returns:
+        Dict mapping gene names (uppercase) to their statistics:
+        {
+            'CD3D': {'avg_log2FC': '2.45', 'pct.1': '0.85', 'pct.2': '0.12', 'p_val_adj': '1.2e-50'},
+            ...
+        }
+    """
+    gene_stats = {}
+
+    # Pattern to match gene statistics table rows
+    # Format: gene_name    number    number    number    number (scientific notation)
+    # Examples:
+    #   IGHA2       7.38  0.89  0.10  0.00e+00
+    #   CD3D       -2.45  0.85  0.12  1.20e-50
+    gene_row_pattern = re.compile(
+        r'^\s*([A-Z][A-Z0-9\-\.]+)\s+(-?\d+\.?\d*)\s+(\d+\.?\d*)\s+(\d+\.?\d*)\s+(\d+\.?\d*[eE][-+]?\d+|\d+\.?\d*)\s*$',
+        re.MULTILINE
+    )
+
+    for msg in conversation_history:
+        content = msg.get('content', '')
+        if isinstance(content, list):
+            content = str(content)
+
+        # Only look at USER messages which contain the gene expression data
+        if msg.get('role', '').upper() == 'USER':
+            matches = gene_row_pattern.findall(content)
+            for match in matches:
+                gene_name, avg_log2fc, pct1, pct2, p_val_adj = match
+                gene_stats[gene_name.upper()] = {
+                    'avg_log2FC': avg_log2fc,
+                    'pct.1': pct1,
+                    'pct.2': pct2,
+                    'p_val_adj': p_val_adj
+                }
+
+    return gene_stats
+
+def create_gene_badge_html(gene: str, gene_stats: Dict[str, Dict[str, str]] = None) -> str:
+    """
+    Create HTML for a gene badge with optional tooltip showing statistics.
+
+    Args:
+        gene: Gene name to display
+        gene_stats: Dictionary of gene statistics (from extract_gene_stats_from_conversation)
+
+    Returns:
+        HTML string for the gene badge with tooltip if stats available
+    """
+    gene_upper = gene.upper().strip()
+
+    if gene_stats and gene_upper in gene_stats:
+        stats = gene_stats[gene_upper]
+        avg_log2fc = stats.get('avg_log2FC', 'N/A')
+        pct1 = stats.get('pct.1', 'N/A')
+        pct2 = stats.get('pct.2', 'N/A')
+        p_val_adj = stats.get('p_val_adj', 'N/A')
+
+        # Determine if log2FC is positive or negative for coloring
+        try:
+            fc_value = float(avg_log2fc)
+            fc_class = 'positive' if fc_value >= 0 else 'negative'
+        except (ValueError, TypeError):
+            fc_class = ''
+
+        tooltip_html = f'''<span class="tooltip">
+            <div class="stat-row"><span class="stat-label">avg_log2FC:</span><span class="stat-value {fc_class}">{avg_log2fc}</span></div>
+            <div class="stat-row"><span class="stat-label">pct.1:</span><span class="stat-value">{pct1}</span></div>
+            <div class="stat-row"><span class="stat-label">pct.2:</span><span class="stat-value">{pct2}</span></div>
+            <div class="stat-row"><span class="stat-label">p_val_adj:</span><span class="stat-value">{p_val_adj}</span></div>
+        </span>'''
+
+        return f'<span class="gene-badge has-stats">{gene}{tooltip_html}</span>'
+    else:
+        # No stats available - add class for styling and native title tooltip
+        return f'<span class="gene-badge no-stats" title="No statistics available">{gene}</span>'
+
 def extract_genes_from_conversation(conversation: str) -> List[str]:
     """
     Extract gene lists from conversation using the check_genes tag.
@@ -1209,6 +1296,9 @@ def generate_summary_report(conversation_history: List[Dict[str, str]], output_f
         str: Path to the saved HTML report
     """
     try:
+        # Extract gene statistics from conversation history for tooltips
+        gene_stats = extract_gene_stats_from_conversation(conversation_history)
+
         # Extract content from conversation history, alternating between assistant and user
         full_conversation = ""
         for msg in conversation_history:
@@ -1352,7 +1442,7 @@ def generate_summary_report(conversation_history: List[Dict[str, str]], output_f
         )
         
         # Convert to HTML and save
-        html_path = format_summary_to_html(summary, output_filename, search_strategy, report_style)
+        html_path = format_summary_to_html(summary, output_filename, search_strategy, report_style, gene_stats=gene_stats)
         print(f"Summary report saved to {html_path}")
         
         # Return the HTML file path
@@ -1598,16 +1688,17 @@ def runCASSIA_annotationboost_additional_task(
         }
 
 def format_summary_to_html(summary_text: str, output_filename: str, search_strategy: str = "breadth", report_style: str = "per_iteration",
-    validator_involvement: str = "v1") -> str:
+    validator_involvement: str = "v1", gene_stats: Dict[str, Dict[str, str]] = None) -> str:
     """
     Convert the tagged summary into a properly formatted HTML report.
-    
+
     Args:
         summary_text: Text with tags like <OVERVIEW>, <ITERATION_1>, etc. or gene-focused tags
         output_filename: Path to save the HTML report
         search_strategy: Search strategy used ("breadth" or "depth")
         report_style: Style of report ("per_iteration" or "total_summary")
-        
+        gene_stats: Dictionary of gene statistics for tooltip display
+
     Returns:
         str: Path to the saved HTML report
     """
@@ -1871,6 +1962,85 @@ def format_summary_to_html(summary_text: str, output_filename: str, search_strat
                     border-radius: 1rem;
                     font-size: 0.9rem;
                     font-weight: 500;
+                    position: relative;
+                    cursor: pointer;
+                    transition: background-color 0.2s;
+                }}
+
+                .gene-badge:hover {{
+                    background-color: #3730a3;
+                }}
+
+                .gene-badge .tooltip {{
+                    visibility: hidden;
+                    opacity: 0;
+                    position: absolute;
+                    bottom: 125%;
+                    left: 50%;
+                    transform: translateX(-50%);
+                    background-color: #1f2937;
+                    color: #f3f4f6;
+                    padding: 0.75rem 1rem;
+                    border-radius: 0.5rem;
+                    font-size: 0.8rem;
+                    font-weight: 400;
+                    white-space: nowrap;
+                    z-index: 1000;
+                    box-shadow: 0 4px 12px rgba(0, 0, 0, 0.25);
+                    transition: visibility 0s, opacity 0.2s;
+                }}
+
+                .gene-badge .tooltip::after {{
+                    content: "";
+                    position: absolute;
+                    top: 100%;
+                    left: 50%;
+                    margin-left: -6px;
+                    border-width: 6px;
+                    border-style: solid;
+                    border-color: #1f2937 transparent transparent transparent;
+                }}
+
+                .gene-badge:hover .tooltip {{
+                    visibility: visible;
+                    opacity: 1;
+                }}
+
+                .gene-badge .tooltip .stat-row {{
+                    display: flex;
+                    justify-content: space-between;
+                    gap: 1rem;
+                    margin-bottom: 0.25rem;
+                }}
+
+                .gene-badge .tooltip .stat-row:last-child {{
+                    margin-bottom: 0;
+                }}
+
+                .gene-badge .tooltip .stat-label {{
+                    color: #9ca3af;
+                    font-size: 0.75rem;
+                }}
+
+                .gene-badge .tooltip .stat-value {{
+                    font-weight: 600;
+                    font-family: ui-monospace, monospace;
+                }}
+
+                .gene-badge .tooltip .stat-value.positive {{
+                    color: #34d399;
+                }}
+
+                .gene-badge .tooltip .stat-value.negative {{
+                    color: #f87171;
+                }}
+
+                .gene-badge.no-stats {{
+                    opacity: 0.7;
+                }}
+
+                .gene-badge.has-stats {{
+                    cursor: help;
                 }}
                 
                 .sub-section {{
@@ -2013,12 +2183,12 @@ def format_summary_to_html(summary_text: str, output_filename: str, search_strat
         if report_style.lower() == "total_summary":
             # Add gene groups for total summary style
             for i, group in enumerate(gene_groups, 1):
-                # Format genes as badges
+                # Format genes as badges with tooltip stats
                 genes = group['genes']
                 gene_badges = ""
                 if genes and genes != "No genes listed":
                     gene_list = [g.strip() for g in re.split(r'[,\s]+', genes) if g.strip()]
-                    gene_badges = '<div class="gene-list">' + ''.join([f'<span class="gene-badge">{gene}</span>' for gene in gene_list]) + '</div>'
+                    gene_badges = '<div class="gene-list">' + ''.join([create_gene_badge_html(gene, gene_stats) for gene in gene_list]) + '</div>'
                 
                 html += f"""
                     <section id="gene-analysis-{i}">
@@ -2040,12 +2210,12 @@ def format_summary_to_html(summary_text: str, output_filename: str, search_strat
         else:
             # Add iterations for per-iteration style
             for iteration in iterations:
-                # Format genes checked as badges
+                # Format genes checked as badges with tooltip stats
                 genes = iteration['genes_checked']
                 gene_badges = ""
                 if genes and genes != "No information available":
                     gene_list = [g.strip() for g in re.split(r'[,\s]+', genes) if g.strip()]
-                    gene_badges = '<div class="gene-list">' + ''.join([f'<span class="gene-badge">{gene}</span>' for gene in gene_list]) + '</div>'
+                    gene_badges = '<div class="gene-list">' + ''.join([create_gene_badge_html(gene, gene_stats) for gene in gene_list]) + '</div>'
                 
                 html += f"""
                     <section id="iteration-{iteration['number']}">

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/agents/annotation_boost/super_annotation_boost.py

@@ -34,7 +34,6 @@ try:
     import seaborn as sns
     SCIENTIFIC_LIBS_AVAILABLE = True
 except ImportError as e:
-    warnings.warn(f"Scientific computing libraries not available: {e}")
     SCIENTIFIC_LIBS_AVAILABLE = False
 
 # Optional imports for enhanced functionality
@@ -43,14 +42,12 @@ try:
     GSEAPY_AVAILABLE = True
 except ImportError:
     GSEAPY_AVAILABLE = False
-    warnings.warn("GSEApy not available - pathway enrichment will use simplified method")
 
 try:
     import requests
     REQUESTS_AVAILABLE = True
 except ImportError:
     REQUESTS_AVAILABLE = False
-    warnings.warn("Requests not available - ontology search will use simplified method")
 
 if not SCIENTIFIC_LIBS_AVAILABLE:
     # Create minimal fallback classes for MVP testing

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/agents/subclustering/subclustering.py

@@ -139,10 +139,10 @@ Remember these subclusters are from a {major_cluster_info} big cluster. You must
 """
 
     # Iterate over each row in the DataFrame
-    for index, row in marker.iterrows():
+    for i, (index, row) in enumerate(marker.iterrows(), start=1):
         cluster_name = row.iloc[0]  # Use iloc for positional indexing
         markers = row.iloc[1]       # Use iloc for positional indexing
-        prompt += f"{index + 1}.{markers}\n"
+        prompt += f"{i}.{markers}\n"
 
     return prompt
 
@@ -384,9 +384,9 @@ def runCASSIA_subclusters(marker, major_cluster_info, output_name,
         from CASSIA.reports.generate_reports import process_evaluation_csv
     except ImportError:
         try:
-            from .generate_reports import process_evaluation_csv
+            from ...reports.generate_reports import process_evaluation_csv
         except ImportError:
-            from generate_reports import process_evaluation_csv
+            from reports.generate_reports import process_evaluation_csv
     import os
     csv_file = output_name if output_name.lower().endswith('.csv') else output_name + '.csv'
     if os.path.exists(csv_file):
@@ -546,9 +546,9 @@ def runCASSIA_n_subcluster(n, marker, major_cluster_info, base_output_name,
         from CASSIA.reports.generate_reports import process_evaluation_csv, create_index_html
     except ImportError:
         try:
-            from .generate_reports import process_evaluation_csv, create_index_html
+            from ...reports.generate_reports import process_evaluation_csv, create_index_html
         except ImportError:
-            from generate_reports import process_evaluation_csv, create_index_html
+            from reports.generate_reports import process_evaluation_csv, create_index_html
     import os
 
     # Generate HTML report for each CSV

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/agents/uncertainty/Uncertainty_quantification.py

@@ -1169,7 +1169,7 @@ def create_and_save_results_dataframe(processed_results, organized_results, outp
     return df
 
 
-def runCASSIA_similarity_score_batch(marker, file_pattern, output_name, celltype_column=None, max_workers=10, model="google/gemini-2.5-flash-preview", provider="openrouter", main_weight=0.5, sub_weight=0.5, temperature=0.0):
+def runCASSIA_similarity_score_batch(marker, file_pattern, output_name, celltype_column=None, max_workers=10, model="google/gemini-2.5-flash-preview", provider="openrouter", main_weight=0.5, sub_weight=0.5, temperature=0.0, generate_report=True, report_output_path=None):
     """
     Process batch results and save them to a CSV file, measuring the time taken.
 
@@ -1184,6 +1184,8 @@ def runCASSIA_similarity_score_batch(marker, file_pattern, output_name, celltype
     main_weight (float): Weight for the main cell type in similarity calculation.
     sub_weight (float): Weight for the sub cell type in similarity calculation.
     temperature (float): Temperature for the LLM calls.
+    generate_report (bool): Whether to generate an HTML report (default: True).
+    report_output_path (str): Path to save the HTML report (default: 'uq_batch_report.html').
     """
 
     # Organize batch results
@@ -1198,13 +1200,30 @@ def runCASSIA_similarity_score_batch(marker, file_pattern, output_name, celltype
 
     # Create and save results dataframe
     create_and_save_results_dataframe(
-        processed_results, 
-        organized_results, 
+        processed_results,
+        organized_results,
         output_name=output_name
     )
 
     print(f"Similarity analysis completed: {output_name}")
 
+    # Generate HTML report if requested
+    if generate_report:
+        try:
+            from CASSIA.reports.generate_report_uncertainty import generate_uq_batch_html_report
+            report_path = report_output_path or 'uq_batch_report.html'
+            generate_uq_batch_html_report(
+                processed_results=processed_results,
+                organized_results=organized_results,
+                output_path=report_path,
+                model=model,
+                provider=provider
+            )
+        except ImportError as e:
+            print(f"Warning: Could not generate report - {e}")
+        except Exception as e:
+            print(f"Warning: Report generation failed - {e}")
+
 
 
 def extract_cell_types_from_results_single(results):
@@ -1322,10 +1341,10 @@ def standardize_cell_types_single(results):
     return ",".join(standardized_results)
 
 
-def runCASSIA_n_times_similarity_score(tissue, species, additional_info, temperature, marker_list, model="google/gemini-2.5-flash-preview", max_workers=10, n=3, provider="openrouter", main_weight=0.5, sub_weight=0.5, validator_involvement="v1", use_reference=False):
+def runCASSIA_n_times_similarity_score(tissue, species, additional_info, temperature, marker_list, model="google/gemini-2.5-flash-preview", max_workers=10, n=3, provider="openrouter", main_weight=0.5, sub_weight=0.5, validator_involvement="v1", use_reference=False, generate_report=True, report_output_path=None):
     """
     Wrapper function for processing cell type analysis using any supported provider.
-    
+
     Args:
         tissue (str): Tissue type
         species (str): Species type
@@ -1338,7 +1357,11 @@ def runCASSIA_n_times_similarity_score(tissue, species, additional_info, tempera
         provider (str): AI provider to use ('openai', 'anthropic', 'openrouter', or a custom URL)
         main_weight (float): Weight for main cell type in similarity calculation
         sub_weight (float): Weight for sub cell type in similarity calculation
-    
+        validator_involvement (str): Validator involvement level
+        use_reference (bool): Whether to use reference information
+        generate_report (bool): Whether to generate an HTML report (default: False)
+        report_output_path (str): Path to save the HTML report (default: 'uq_report.html')
+
     Returns:
         dict: Analysis results including consensus types, cell types, and scores
     """
@@ -1384,8 +1407,8 @@ Output in JSON format:
     # Calculate similarity score
     parsed_results = parse_results_to_dict_single(results)
     consensus_score, consensus_1, consensus_2 = consensus_similarity_flexible_single(parsed_results,main_weight=main_weight,sub_weight=sub_weight)
-    
-    return {
+
+    final_results = {
         'unified_results': standardized_results,
         'consensus_types': (consensus_1, consensus_2),
         'general_celltype_llm': general_celltype,
@@ -1397,3 +1420,24 @@ Output in JSON format:
         'original_results': results
     }
 
+    # Generate HTML report if requested
+    if generate_report:
+        try:
+            from CASSIA.reports.generate_report_uncertainty import generate_uq_html_report
+            report_path = report_output_path or 'uq_report.html'
+            generate_uq_html_report(
+                results=final_results,
+                output_path=report_path,
+                tissue=tissue,
+                species=species,
+                model=model,
+                n_iterations=n,
+                marker_list=marker_list
+            )
+        except ImportError as e:
+            print(f"Warning: Could not generate report - {e}")
+        except Exception as e:
+            print(f"Warning: Report generation failed - {e}")
+
+    return final_results
+

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/comparison/symphony_compare.py

@@ -28,13 +28,13 @@ def symphonyCompare(
     celltypes: List[str],
     marker_set: str,
     species: str = "human",
-    model_preset: str = "symphony",
+    model_preset: str = "budget",
     custom_models: Optional[List[str]] = None,
     output_dir: Optional[str] = None,
     output_basename: Optional[str] = None,
     enable_discussion: bool = True,
-    max_discussion_rounds: int = 2,
-    consensus_threshold: float = 0.8,
+    max_discussion_rounds: int = 3,
+    consensus_threshold: float = 2/3,
     generate_report: bool = True,
     api_key: Optional[str] = None,
     verbose: bool = True
@@ -51,17 +51,16 @@ def symphonyCompare(
         celltypes (List[str]): List of 2-4 cell types to compare
         marker_set (str): Comma-separated string of gene markers to analyze
         species (str): Species being analyzed (default: "human")
-        model_preset (str): Preset model configuration. Options:
-            - "symphony": High-performance ensemble (Claude, GPT-4, Gemini Pro)
-            - "quartet": Balanced 4-model ensemble
-            - "budget": Cost-effective models
+        model_preset (str): Preset model configuration (default: "budget"). Options:
+            - "budget": Cost-effective models (DeepSeek, Grok 4 Fast, Kimi K2, Gemini Flash)
+            - "premium": High-performance ensemble (Gemini 3 Pro, Claude Sonnet 4.5, GPT-5.1, Grok 4)
             - "custom": Use custom_models list
         custom_models (List[str]): Custom list of models to use (when model_preset="custom")
         output_dir (str): Directory to save results (default: current directory)
         output_basename (str): Base name for output files (auto-generated if None)
         enable_discussion (bool): Enable automatic discussion rounds when no consensus (default: True)
-        max_discussion_rounds (int): Maximum discussion rounds to perform (default: 2)
-        consensus_threshold (float): Fraction of models that must agree for consensus (default: 0.8)
+        max_discussion_rounds (int): Maximum discussion rounds to perform (default: 3)
+        consensus_threshold (float): Fraction of models that must agree for consensus (default: 2/3)
         generate_report (bool): Generate interactive HTML report (default: True)
         api_key (str): OpenRouter API key (uses environment variable if None)
         verbose (bool): Print progress messages (default: True)
@@ -125,7 +124,14 @@ def symphonyCompare(
     if api_key is None:
         api_key = os.environ.get('OPENROUTER_API_KEY')
     if not api_key:
-        raise ValueError("OPENROUTER_API_KEY not found. Set it as an environment variable or pass it as api_key parameter.")
+        raise ValueError(
+            "OPENROUTER_API_KEY not found.\n"
+            "Symphony Compare uses models from multiple providers (OpenAI, Anthropic, Google, etc.),\n"
+            "so an OpenRouter API key is required to access all models through a unified endpoint.\n"
+            "Get your API key at: https://openrouter.ai/keys\n"
+            "Then set it as an environment variable: export OPENROUTER_API_KEY='your-key'\n"
+            "Or pass it directly: symphonyCompare(..., api_key='your-key')"
+        )
     
     # Input validation
     if not celltypes or len(celltypes) < 2 or len(celltypes) > 4:
@@ -144,36 +150,49 @@ def symphonyCompare(
     csv_file = os.path.join(output_dir, f"{output_basename}.csv")
     html_file = os.path.join(output_dir, f"{output_basename}_report.html") if generate_report else None
     
-    # Define model presets
-    model_presets = {
-        "symphony": [
-            "anthropic/claude-3.7-sonnet",
-            "openai/o4-mini-high",
-            "google/gemini-2.5-pro-preview"
-        ],
-        "quartet": [
-            "anthropic/claude-3.7-sonnet",
-            "openai/o4-mini-high", 
-            "google/gemini-2.5-pro-preview",
-            "meta-llama/llama-3.3-405b"
+    # Load model presets and personas from JSON config file
+    config_file = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'model_config.json')
+
+    # Default fallback configuration
+    default_presets = {
+        "premium": [
+            "google/gemini-3-pro-preview",
+            "anthropic/claude-sonnet-4.5",
+            "openai/gpt-5.1",
+            "x-ai/grok-4"
         ],
         "budget": [
-            "google/gemini-2.5-flash",
-            "deepseek/deepseek-chat-v3-0324",
-            "x-ai/grok-3-mini-beta"
+            "deepseek/deepseek-v3.2",
+            "x-ai/grok-4-fast",
+            "moonshotai/kimi-k2-thinking",
+            "google/gemini-2.5-flash"
         ]
     }
-    
-    # Researcher persona names for each model
-    model_personas = {
-        "google/gemini-2.5-flash": "Dr. Ada Lovelace",
-        "deepseek/deepseek-chat-v3-0324": "Dr. Alan Turing", 
-        "x-ai/grok-3-mini-beta": "Dr. Marie Curie",
-        "anthropic/claude-3.7-sonnet": "Dr. Claude Shannon",
-        "openai/o4-mini-high": "Dr. Albert Einstein",
-        "google/gemini-2.5-pro-preview": "Dr. Emmy Noether",
-        "meta-llama/llama-3.3-405b": "Dr. Rosalind Franklin"
+
+    default_personas = {
+        "google/gemini-3-pro-preview": "Dr. Emmy Noether",
+        "anthropic/claude-sonnet-4.5": "Dr. Claude Shannon",
+        "openai/gpt-5.1": "Dr. Albert Einstein",
+        "x-ai/grok-4": "Dr. Marie Curie",
+        "deepseek/deepseek-v3.2": "Dr. Alan Turing",
+        "x-ai/grok-4-fast": "Dr. Nikola Tesla",
+        "moonshotai/kimi-k2-thinking": "Dr. Ada Lovelace",
+        "google/gemini-2.5-flash": "Dr. Rosalind Franklin"
     }
+
+    # Try to load from JSON config file
+    try:
+        with open(config_file, 'r') as f:
+            config = json.load(f)
+        model_presets = config.get('presets', default_presets)
+        model_personas = config.get('personas', default_personas)
+        if verbose:
+            print(f"  Loaded model configuration from: {config_file}")
+    except (FileNotFoundError, json.JSONDecodeError) as e:
+        model_presets = default_presets
+        model_personas = default_personas
+        if verbose:
+            print(f"  Using default model configuration (config file not found or invalid)")
     
     # Select models based on preset or custom list
     if model_preset == "custom" and custom_models:
@@ -182,8 +201,8 @@ def symphonyCompare(
         model_list = model_presets[model_preset]
     else:
         if verbose:
-            print(f"Warning: Unknown preset '{model_preset}'. Using 'symphony' preset.")
-        model_list = model_presets["symphony"]
+            print(f"Warning: Unknown preset '{model_preset}'. Using 'budget' preset.")
+        model_list = model_presets["budget"]
     
     # Get persona names
     model_to_persona = {m: model_personas.get(m, f"Researcher_{m.split('/')[-1]}") for m in model_list}
@@ -197,6 +216,8 @@ def symphonyCompare(
         print(f"🤖 Models: {', '.join([model_to_persona[m].split()[-1] for m in model_list])}")
         if enable_discussion:
             print(f"💬 Discussion: Enabled (max {max_discussion_rounds} rounds)")
+        if model_preset == "budget":
+            print(f"💡 Tip: For better performance, use model_preset='premium'")
         print(f"{'='*60}\n")
     
     # Construct initial prompt

{cassia-0.3.1.dev5 → cassia-0.3.3}/CASSIA/reports/__init__.py

@@ -9,6 +9,7 @@ from .generate_reports import (
     runCASSIA_generate_score_report
 )
 from .generate_hypothesis_report import create_html_report
+from .generate_report_uncertainty import generate_uq_html_report
 
 # Alias for backward compatibility
 generate_hypothesis_report = create_html_report
@@ -21,4 +22,5 @@ __all__ = [
     'runCASSIA_generate_score_report',
     'create_html_report',
     'generate_hypothesis_report',
+    'generate_uq_html_report',
 ]