simple_tool_comparison.py

# Simple comparison: All tools (regular embeddings) + Kompot (both flavors)
from gather_stability_columns import gather_stability_columns, plot_multi_tool_correlation_vs_parameter
import os

# Create plots directory
os.makedirs('plots', exist_ok=True)

def load_tool_data(directory, column_name, table_type, tool_name):
    """Load tool data with error handling."""
    try:
        print(f"Loading {tool_name}...")
        df, _ = gather_stability_columns(directory, column_name, table_type)
        print(f"✓ {tool_name}: {df.shape[0]} genes/cells, {df.shape[1]} samples")
        return df
    except Exception as e:
        print(f"✗ {tool_name}: {e}")
        return None

# Define tool configurations
tools_config = [
    ('data/processed/aging_scanpy_subsampling', 'scanpy_de_log_fold_change_Young_to_Old', 'var', 'Scanpy'),
    ('data/processed/aging_mrvi_subsampling', 'mrvi_de_log_fold_change_Young_to_Old', 'var', 'mrVI'),
    ('data/processed/aging_lemur_subsampling', 'lemur_de_log_fold_change_Young_to_Old', 'var', 'LEMUR'),
    ('data/processed/aging_seurat_subsampling', 'seurat_de_log_fold_change_Young_to_Old', 'var', 'Seurat'),
    ('data/processed/aging_milo_subsampling', 'milo_da_log_fold_change_Young_to_Old', 'obs', 'Milo'),
    ('data/processed/aging_meld_subsampling', 'meld_da_log_fold_change_Young_to_Old', 'obs', 'MELD'),
    ('data/processed/aging_deseq2_subsampling', 'deseq2_de_log_fold_change_Young_to_Old', 'var', 'DESeq2'),
]

# Load all available tools
tool_data = {}
print("Loading tools with regular embeddings...")
for directory, column_name, table_type, tool_name in tools_config:
    df = load_tool_data(directory, column_name, table_type, tool_name)
    if df is not None:
        tool_data[tool_name] = df

# Load Kompot with both embedding strategies
print("\nLoading Kompot with both embedding strategies...")
kompot_regular = load_tool_data('data/processed/aging_subsampling', 'kompot_de_mahalanobis_Young_to_Old', 'var', 'Kompot (Regular)')
kompot_precomputed = load_tool_data('data/processed/aging_precomputed_embeddings_subsampling', 'kompot_de_mahalanobis_Young_to_Old', 'var', 'Kompot (Precomputed)')

if kompot_regular is not None:
    tool_data['Kompot (Regular)'] = kompot_regular
if kompot_precomputed is not None:
    tool_data['Kompot (Precomputed)'] = kompot_precomputed

# Create comparison plot if we have data
if tool_data:
    print(f"\nCreating plot with {len(tool_data)} tools...")
    fig = plot_multi_tool_correlation_vs_parameter(
        tool_data=tool_data,
        correlation_method='spearman',
        title='DE/DA Tool Comparison: Regular Embeddings + Kompot Both Flavors',
        reference_index=0,
        save_path='plots/all_tools_kompot_both_flavors.png'
    )
    print("✓ Saved: plots/all_tools_kompot_both_flavors.png")
else:
    print("No tool data loaded successfully.")