Config rewrite

CONFIG_IMPROVEMENT.md

@@ -1,176 +0,0 @@
-# MuMDIA Configuration Management Improvement
-
-## The Problem
-The original `run.py` had extremely complex argument parsing and configuration management:
-
-- **~100 lines** of argument parsing code
-- **Complex merging logic** between CLI args and config files  
-- **Difficult to maintain** argument parsing functions
-- **Hard to understand** configuration flow
-- **Manual config validation** and error handling
-
-## The Solution: Simplified Configuration with Dataclasses
-
-### Before (Complex approach in `run.py`):
-
-```python
-def parse_arguments() -> Tuple[argparse.ArgumentParser, argparse.Namespace]:
-    """Parse command line arguments - 50+ lines of argparse setup"""
-    parser = argparse.ArgumentParser()
-    parser.add_argument("--mzml_file", type=str, help="Path to mzML file", default=None)
-    parser.add_argument("--fasta_file", type=str, help="Path to FASTA file", default=None)
-    # ... 30+ more arguments ...
-    args = parser.parse_args()
-    return parser, args
-
-def was_arg_explicitly_provided(parser: argparse.ArgumentParser, arg_name: str) -> bool:
-    """Check if argument was explicitly provided - complex logic"""
-    for action in parser._actions:
-        if arg_name in action.dest:
-            for option in action.option_strings:
-                if option in sys.argv:
-                    return True
-    return False
-
-def modify_config(parser: argparse.ArgumentParser, args: argparse.Namespace, config_path: str) -> Dict[str, Any]:
-    """Load and modify config - 50+ lines of complex merging logic"""
-    # Load base config from JSON
-    # Override with CLI args using complex checking
-    # Save effective config
-    # Return merged config dict
-
-# In main():
-parser, args = parse_arguments()
-config = modify_config(parser, args, args.config_file)
-# Extract all individual values from config dict
-mzml_file = config["mzml_file"]
-fasta_file = config["fasta_file"]
-# ... dozens more extractions ...
-```
-
-### After (Clean dataclass approach in `config_manager_clean.py`):
-
-```python
-@dataclass
-class MuMDIAConfig:
-    """Clean, type-safe configuration with defaults"""
-    mzml_file: str = ""
-    fasta_file: str = ""
-    result_dir: str = "results"
-    n_windows: int = 10
-    training_fdr: float = 0.05
-    final_fdr: float = 0.01
-    model_type: str = "xgboost"
-    no_cache: bool = False
-    clean: bool = False
-    sage_only: bool = False
-    # ... all options clearly defined
-
-    @classmethod
-    def from_json(cls, json_path: str) -> "MuMDIAConfig":
-        """Simple JSON loading with error handling"""
-
-    @classmethod  
-    def from_args(cls, args=None) -> "MuMDIAConfig":
-        """Simple CLI parsing with config file support"""
-
-    def validate(self) -> None:
-        """Clean validation logic"""
-
-def get_config() -> MuMDIAConfig:
-    """One-liner to get validated config"""
-    config = MuMDIAConfig.from_args()
-    config.validate()
-    return config
-
-# In main():
-config = get_config()  # That's it! One line replaces 100+ lines!
-# Direct attribute access:
-print(config.mzml_file)
-print(config.n_windows)
-```
-
-## Key Improvements
-
-### 1. **Dramatic Code Reduction**
-- **Before**: ~100 lines of complex parsing logic
-- **After**: 1 line: `config = get_config()`
-- **Reduction**: 99% fewer lines for config management!
-
-### 2. **Type Safety** 
-- **Before**: Untyped dictionary access like `config["mzml_file"]`
-- **After**: Type-safe attribute access like `config.mzml_file`
-- **Benefit**: IDE autocomplete, type checking, fewer runtime errors
-
-### 3. **Clear Defaults**
-- **Before**: Defaults scattered across argparse definitions
-- **After**: All defaults clearly visible in dataclass definition
-- **Benefit**: Easy to see and modify default values
-
-### 4. **Better Error Handling**
-- **Before**: Manual validation scattered throughout code
-- **After**: Centralized validation in `validate()` method
-- **Benefit**: Consistent error messages and validation logic
-
-### 5. **Simpler Usage Patterns**
-
-#### Command line usage:
-```bash
-# Simple usage
-python run_simple.py --mzml_file data.mzML --fasta_file proteins.fasta
-
-# With options
-python run_simple.py --mzml_file data.mzML --fasta_file proteins.fasta --n_windows 5 --verbose --no-cache
-
-# With config file
-python run_simple.py --config_file my_config.json
-
-# Config file with CLI overrides 
-python run_simple.py --config_file my_config.json --clean --result_dir custom_results
-```
-
-#### JSON config files:
-```json
-{
-  "mzml_file": "data.mzML",
-  "fasta_file": "proteins.fasta", 
-  "result_dir": "my_results",
-  "n_windows": 15,
-  "training_fdr": 0.1,
-  "model_type": "nn",
-  "verbose": true
-}
-```
-
-## Implementation Status
-
-✅ **Created**: `config_manager_clean.py` - Complete simplified config system
-✅ **Created**: `run_simple.py` - Demo of clean config usage
-✅ **Created**: `config_demo.py` - Working demonstration
-✅ **Tested**: Both CLI arguments and JSON config files work perfectly
-
-## Migration Path
-
-The new config system is **fully backwards compatible**. You can:
-
-1. **Immediate adoption**: Use `config_manager_clean.py` for new features
-2. **Gradual migration**: Replace complex config logic piece by piece  
-3. **Side-by-side**: Run both systems during transition period
-
-## Developer Experience Impact
-
-**Before**: Developers had to:
-- Navigate 100+ lines of complex argument parsing
-- Understand config merging logic  
-- Manually handle validation
-- Debug dictionary key errors
-- Maintain scattered default values
-
-**After**: Developers can:
-- See all configuration at a glance in the dataclass
-- Get IDE support with autocomplete and type checking
-- Add new config options by just adding a dataclass field
-- Trust that validation is centralized and consistent
-- Focus on business logic instead of configuration plumbing
-
-The new approach makes MuMDIA **much more maintainable and developer-friendly**!