Upload data_retrieval.py with huggingface_hub

data_retrieval.py

@@ -0,0 +1,682 @@
+#!/usr/bin/env python3
+"""
+Phase 1: Data Retrieval & Setup
+Aging Fly Cell Atlas (AFCA) - GSE218661
+
+This script programmatically retrieves h5ad files and metadata from GSE218661
+for the Aging Fly Cell Atlas study. Downloads both head and body data files.
+
+Key features:
+- Downloads h5ad files from GEO supplementary files 
+- Extracts comprehensive metadata from all available sources
+- Organizes data in proper directory structure
+- Validates downloaded files
+"""
+
+import os
+import sys
+import requests
+import GEOparse
+import pandas as pd
+import json
+import warnings
+import gzip
+import scanpy as sc
+from pathlib import Path
+from typing import Dict, List, Optional, Tuple
+import time
+from urllib.parse import urlparse
+import hashlib
+
+# Suppress warnings for cleaner output
+warnings.filterwarnings('ignore')
+
+def setup_directories() -> Dict[str, Path]:
+    """Create necessary directory structure for AFCA data."""
+    
+    print("🗂️  SETTING UP DIRECTORY STRUCTURE")
+    print("=" * 50)
+    
+    # Define directory structure
+    dirs = {
+        'data': Path('data'),
+        'raw': Path('data/raw'),
+        'processed': Path('processed'), 
+        'metadata': Path('data/metadata'),
+        'logs': Path('data/logs'),
+        'supplementary': Path('data/raw/supplementary')
+    }
+    
+    # Create directories
+    for name, path in dirs.items():
+        path.mkdir(exist_ok=True, parents=True)
+        print(f"   ✅ Created: {path}")
+    
+    return dirs
+
+def extract_geo_metadata(accession: str = "GSE218661") -> Dict:
+    """Extract comprehensive metadata from GEO using GEOparse."""
+    
+    print(f"\n📊 EXTRACTING GEO METADATA FOR {accession}")
+    print("=" * 50)
+    
+    try:
+        # Download GEO metadata
+        print(f"   📡 Connecting to GEO database...")
+        gse = GEOparse.get_GEO(geo=accession, destdir="data/metadata/")
+        
+        metadata = {
+            'accession': accession,
+            'title': gse.metadata.get('title', [''])[0],
+            'summary': gse.metadata.get('summary', [''])[0],
+            'overall_design': gse.metadata.get('overall_design', [''])[0],
+            'submission_date': gse.metadata.get('submission_date', [''])[0],
+            'last_update_date': gse.metadata.get('last_update_date', [''])[0],
+            'organism': gse.metadata.get('organism', []),
+            'platform_organism': gse.metadata.get('platform_organism', []),
+            'contact_email': gse.metadata.get('contact_email', [''])[0],
+            'contact_name': gse.metadata.get('contact_name', [''])[0],
+            'contact_institute': gse.metadata.get('contact_institute', [''])[0],
+            'supplementary_file': gse.metadata.get('supplementary_file', []),
+            'relation': gse.metadata.get('relation', []),
+            'sample_count': len(gse.gsms),
+            'platform_count': len(gse.gpls),
+            'samples': {},
+            'platforms': {}
+        }
+        
+        # Extract sample metadata
+        print(f"   🧪 Extracting metadata for {len(gse.gsms)} samples...")
+        for gsm_name, gsm in gse.gsms.items():
+            metadata['samples'][gsm_name] = {
+                'title': gsm.metadata.get('title', [''])[0],
+                'source_name_ch1': gsm.metadata.get('source_name_ch1', [''])[0],
+                'organism_ch1': gsm.metadata.get('organism_ch1', [''])[0],
+                'characteristics_ch1': gsm.metadata.get('characteristics_ch1', []),
+                'treatment_protocol_ch1': gsm.metadata.get('treatment_protocol_ch1', [''])[0],
+                'extract_protocol_ch1': gsm.metadata.get('extract_protocol_ch1', [''])[0],
+                'description': gsm.metadata.get('description', [''])[0],
+                'data_processing': gsm.metadata.get('data_processing', []),
+                'platform_id': gsm.metadata.get('platform_id', [''])[0],
+                'contact_name': gsm.metadata.get('contact_name', [''])[0],
+                'supplementary_file': gsm.metadata.get('supplementary_file', []),
+                'submission_date': gsm.metadata.get('submission_date', [''])[0],
+                'last_update_date': gsm.metadata.get('last_update_date', [''])[0]
+            }
+        
+        # Extract platform metadata
+        print(f"   🔬 Extracting metadata for {len(gse.gpls)} platforms...")
+        for gpl_name, gpl in gse.gpls.items():
+            metadata['platforms'][gpl_name] = {
+                'title': gpl.metadata.get('title', [''])[0],
+                'organism': gpl.metadata.get('organism', [''])[0],
+                'technology': gpl.metadata.get('technology', [''])[0],
+                'distribution': gpl.metadata.get('distribution', [''])[0],
+                'description': gpl.metadata.get('description', [''])[0],
+                'submission_date': gpl.metadata.get('submission_date', [''])[0],
+                'last_update_date': gpl.metadata.get('last_update_date', [''])[0]
+            }
+        
+        print(f"   ✅ Successfully extracted metadata for {accession}")
+        return metadata, gse
+        
+    except Exception as e:
+        print(f"   ❌ Error extracting GEO metadata: {e}")
+        return {}, None
+
+def download_geo_supplementary_files(gse, dirs: Dict[str, Path]) -> Dict[str, bool]:
+    """Download supplementary files from GEO which should contain h5ad files."""
+    
+    print("\n📦 DOWNLOADING GEO SUPPLEMENTARY FILES")
+    print("=" * 50)
+    
+    supp_dir = dirs['supplementary']
+    download_results = {'supplementary_files': False}
+    
+    try:
+        print(f"   📂 Downloading supplementary files to: {supp_dir}")
+        
+        # Check if files already exist
+        existing_files = list(supp_dir.glob('*'))
+        if existing_files:
+            print(f"   ✅ Found {len(existing_files)} existing files in {supp_dir}")
+            download_results['supplementary_files'] = True
+        else:
+            # Download supplementary files
+            gse.download_supplementary_files(directory=str(supp_dir))
+            
+            # Check if download was successful
+            downloaded_files = list(supp_dir.glob('*'))
+            if downloaded_files:
+                print(f"   ✅ Successfully downloaded {len(downloaded_files)} supplementary files")
+                download_results['supplementary_files'] = True
+            else:
+                print(f"   ❌ No supplementary files downloaded")
+                
+    except Exception as e:
+        print(f"   ❌ Error downloading supplementary files: {e}")
+        print(f"   🔍 You may need to manually download files from:")
+        print(f"       https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218661")
+    
+    return download_results
+
+def download_h5ad_files_manually(gse, dirs: Dict[str, Path]) -> Dict[str, bool]:
+    """Manually download h5ad files using URLs extracted from GEO metadata."""
+    
+    print("\n📥 EXTRACTING H5AD URLS FROM GEO AND DOWNLOADING")
+    print("=" * 50)
+    
+    supp_dir = dirs['supplementary']
+    download_results = {'h5ad_head': False, 'h5ad_body': False}
+    
+    # Extract supplementary file URLs from GEO metadata
+    h5ad_files = {}
+    
+    # Check GSE-level supplementary files
+    if hasattr(gse, 'metadata') and 'supplementary_file' in gse.metadata:
+        supp_files = gse.metadata['supplementary_file']
+        print(f"   🔍 Found {len(supp_files)} GSE-level supplementary files")
+        
+        for supp_file in supp_files:
+            if '.h5ad' in supp_file.lower():
+                print(f"      📄 H5AD file found: {supp_file}")
+                
+                # Determine tissue type from filename
+                if 'head' in supp_file.lower():
+                    tissue = 'head'
+                elif 'body' in supp_file.lower():
+                    tissue = 'body'
+                else:
+                    tissue = 'unknown'
+                
+                # Extract filename from URL
+                filename = supp_file.split('/')[-1]
+                
+                # Convert FTP URLs to HTTP URLs for requests compatibility
+                url = supp_file
+                if url.startswith('ftp://ftp.ncbi.nlm.nih.gov'):
+                    url = url.replace('ftp://ftp.ncbi.nlm.nih.gov', 'https://ftp.ncbi.nlm.nih.gov')
+                
+                h5ad_files[tissue] = {
+                    'url': url,
+                    'filename': filename
+                }
+    
+    # Also check individual GSM samples for supplementary files
+    for gsm_name, gsm in gse.gsms.items():
+        if hasattr(gsm, 'metadata') and 'supplementary_file' in gsm.metadata:
+            supp_files = gsm.metadata['supplementary_file']
+            for supp_file in supp_files:
+                if '.h5ad' in supp_file.lower():
+                    print(f"      📄 GSM H5AD file found in {gsm_name}: {supp_file}")
+                    
+                    # Determine tissue type from filename or GSM metadata
+                    tissue = 'unknown'
+                    if 'head' in supp_file.lower():
+                        tissue = 'head'
+                    elif 'body' in supp_file.lower():
+                        tissue = 'body'
+                    elif hasattr(gsm, 'metadata') and 'source_name_ch1' in gsm.metadata:
+                        source = gsm.metadata['source_name_ch1'][0].lower()
+                        if 'head' in source:
+                            tissue = 'head'
+                        elif 'body' in source:
+                            tissue = 'body'
+                    
+                    filename = supp_file.split('/')[-1]
+                    
+                    # Only add if we don't already have this tissue or if this looks more comprehensive
+                    if tissue not in h5ad_files or 'combined' in filename.lower():
+                        # Convert FTP URLs to HTTP URLs for requests compatibility
+                        url = supp_file
+                        if url.startswith('ftp://ftp.ncbi.nlm.nih.gov'):
+                            url = url.replace('ftp://ftp.ncbi.nlm.nih.gov', 'https://ftp.ncbi.nlm.nih.gov')
+                        
+                        h5ad_files[tissue] = {
+                            'url': url,
+                            'filename': filename
+                        }
+    
+    # If no h5ad files found in metadata, construct URLs based on GEO conventions
+    if not h5ad_files:
+        print("   ⚠️  No h5ad files found in GEO metadata, constructing standard URLs...")
+        accession = gse.get_accession()
+        base_url = f"https://ftp.ncbi.nlm.nih.gov/geo/series/{accession[:-3]}nnn/{accession}/suppl/"
+        
+        h5ad_files = {
+            'head': {
+                'url': f"{base_url}{accession}_adata_head_S_v1.0.h5ad.gz",
+                'filename': f"{accession}_adata_head_S_v1.0.h5ad.gz"
+            },
+            'body': {
+                'url': f"{base_url}{accession}_adata_body_S_v1.0.h5ad.gz", 
+                'filename': f"{accession}_adata_body_S_v1.0.h5ad.gz"
+            }
+        }
+        print(f"   🔧 Constructed URLs for {accession}")
+    
+    print(f"\n   📋 H5AD files to download:")
+    for tissue, file_info in h5ad_files.items():
+        print(f"      🧬 {tissue.title()}: {file_info['filename']}")
+        print(f"         URL: {file_info['url']}")
+    
+    # Download each h5ad file
+    for tissue, file_info in h5ad_files.items():
+        file_path = supp_dir / file_info['filename']
+        
+        if file_path.exists():
+            print(f"\n   ✅ {tissue.title()} h5ad file already exists: {file_path}")
+            download_results[f'h5ad_{tissue}'] = True
+            continue
+            
+        try:
+            print(f"\n   📡 Downloading {tissue} h5ad file...")
+            print(f"      Source: {file_info['url']}")
+            print(f"      Destination: {file_path}")
+            
+            response = requests.get(file_info['url'], stream=True)
+            response.raise_for_status()
+            
+            # Get file size for progress tracking
+            total_size = int(response.headers.get('content-length', 0))
+            
+            with open(file_path, 'wb') as f:
+                downloaded = 0
+                for chunk in response.iter_content(chunk_size=8192):
+                    if chunk:
+                        f.write(chunk)
+                        downloaded += len(chunk)
+                        if total_size > 0:
+                            percent = (downloaded / total_size) * 100
+                            print(f"\r      Progress: {percent:.1f}% ({downloaded:,}/{total_size:,} bytes)", end='', flush=True)
+            
+            print(f"\n   ✅ Successfully downloaded {tissue} h5ad file: {file_path}")
+            download_results[f'h5ad_{tissue}'] = True
+            
+        except Exception as e:
+            print(f"\n   ❌ Error downloading {tissue} h5ad file: {e}")
+            continue
+    
+    return download_results
+
+def process_h5ad_files(dirs: Dict[str, Path]) -> Dict[str, any]:
+    """Process h5ad files and extract information."""
+    
+    print("\n🔬 PROCESSING H5AD FILES")
+    print("=" * 50)
+    
+    supp_dir = dirs['supplementary']
+    data_dir = dirs['data']
+    h5ad_info = {
+        'files_found': [],
+        'files_processed': {},
+        'total_cells': 0,
+        'total_genes': 0,
+        'datasets': {}
+    }
+    
+    # Find h5ad files
+    h5ad_files = list(supp_dir.glob('*.h5ad'))
+    if not h5ad_files:
+        # Also check for compressed files
+        h5ad_files.extend(list(supp_dir.glob('*.h5ad.gz')))
+    
+    print(f"   🔍 Found {len(h5ad_files)} h5ad files")
+    
+    for h5ad_file in h5ad_files:
+        try:
+            print(f"   📖 Processing: {h5ad_file.name}")
+            
+            # Read h5ad file
+            if h5ad_file.suffix == '.gz':
+                # Handle compressed files
+                print(f"      📂 Decompressing {h5ad_file.name}")
+                with gzip.open(h5ad_file, 'rb') as f_in:
+                    decompressed_file = h5ad_file.with_suffix('')
+                    with open(decompressed_file, 'wb') as f_out:
+                        f_out.write(f_in.read())
+                adata = sc.read_h5ad(decompressed_file)
+                
+                # Move to data root with clean name and remove compressed file
+                tissue_type = 'unknown'
+                filename_lower = h5ad_file.name.lower()
+                if 'head' in filename_lower:
+                    tissue_type = 'head'
+                elif 'body' in filename_lower:
+                    tissue_type = 'body'
+                
+                final_filename = f"afca_{tissue_type}.h5ad"
+                final_path = data_dir / final_filename
+                
+                print(f"      📁 Moving to data root: {final_path}")
+                decompressed_file.rename(final_path)
+                
+                print(f"      🗑️  Removing compressed file: {h5ad_file}")
+                h5ad_file.unlink()
+                
+                # Update file reference for processing
+                h5ad_file = final_path
+                
+            else:
+                adata = sc.read_h5ad(h5ad_file)
+            
+            # Extract basic information
+            file_info = {
+                'filename': h5ad_file.name,
+                'filepath': str(h5ad_file),
+                'n_obs': adata.n_obs,
+                'n_vars': adata.n_vars,
+                'obs_columns': list(adata.obs.columns),
+                'var_columns': list(adata.var.columns),
+                'uns_keys': list(adata.uns.keys()) if adata.uns else [],
+                'obsm_keys': list(adata.obsm.keys()) if adata.obsm else [],
+                'varm_keys': list(adata.varm.keys()) if adata.varm else [],
+            }
+            
+            # Identify tissue type from filename or metadata
+            tissue_type = 'unknown'
+            filename_lower = h5ad_file.name.lower()
+            if 'head' in filename_lower:
+                tissue_type = 'head'
+            elif 'body' in filename_lower:
+                tissue_type = 'body'
+            elif 'combined' in filename_lower or 'full' in filename_lower:
+                tissue_type = 'combined'
+            
+            file_info['tissue_type'] = tissue_type
+            
+            # Extract age information if available
+            if 'age' in adata.obs.columns:
+                ages = adata.obs['age'].unique()
+                file_info['ages'] = list(ages)
+                print(f"      📅 Ages found: {ages}")
+            
+            # Extract cell type information if available
+            cell_type_cols = [col for col in adata.obs.columns 
+                            if any(term in col.lower() for term in ['cell_type', 'celltype', 'annotation', 'cluster'])]
+            if cell_type_cols:
+                file_info['cell_type_columns'] = cell_type_cols
+                for col in cell_type_cols[:2]:  # Limit to first 2 to avoid too much output
+                    cell_types = adata.obs[col].unique()
+                    file_info[f'{col}_unique_values'] = len(cell_types)
+                    print(f"      🧬 {col}: {len(cell_types)} unique values")
+            
+            h5ad_info['files_processed'][h5ad_file.name] = file_info
+            h5ad_info['total_cells'] += adata.n_obs
+            h5ad_info['total_genes'] = max(h5ad_info['total_genes'], adata.n_vars)
+            
+            print(f"      ✅ Processed: {adata.n_obs:,} cells × {adata.n_vars:,} genes")
+            
+        except Exception as e:
+            print(f"      ❌ Error processing {h5ad_file.name}: {e}")
+            continue
+    
+    h5ad_info['files_found'] = [f.name for f in h5ad_files]
+    
+    if h5ad_info['files_processed']:
+        print(f"\n   📊 Summary:")
+        print(f"      📁 Files processed: {len(h5ad_info['files_processed'])}")
+        print(f"      🧬 Total cells: {h5ad_info['total_cells']:,}")
+        print(f"      🧮 Max genes: {h5ad_info['total_genes']:,}")
+        print(f"      📂 Final h5ad files location: data/")
+    
+    return h5ad_info
+
+def create_afca_data_info() -> Dict:
+    """Create comprehensive information about AFCA dataset."""
+    
+    print("\n📋 CREATING AFCA DATA INFORMATION")
+    print("=" * 50)
+    
+    afca_info = {
+        'dataset_name': 'Aging Fly Cell Atlas (AFCA)',
+        'accession': 'GSE218661',
+        'publication': {
+            'title': 'Aging Fly Cell Atlas identifies exhaustive aging features at cellular resolution',
+            'authors': 'Lu, T.-C., Brbić, M., Park, Y.-J., et al.',
+            'journal': 'Science',
+            'year': 2023,
+            'volume': 380,
+            'issue': 6650,
+            'doi': '10.1126/science.adg0934'
+        },
+        'data_description': {
+            'organism': 'Drosophila melanogaster',
+            'technology': '10x Chromium single-nucleus RNA-seq (snRNA-seq)',
+            'total_nuclei': '868,000+',
+            'cell_types': 163,
+            'ages': ['5d', '30d', '50d', '70d'],
+            'sexes': ['Male', 'Female'],
+            'tissues': ['Head', 'Body']
+        },
+        'data_access': {
+            'web_portal': 'https://hongjielilab.org/afca/',
+            'geo_repository': 'https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218661',
+            'zenodo': 'https://doi.org/10.5281/zenodo.7853649',
+            'cellxgene_head': 'https://cellxgene.cziscience.com/',
+            'cellxgene_body': 'https://cellxgene.cziscience.com/',  
+            'cellxgene_combined': 'https://cellxgene.cziscience.com/'
+        }
+    }
+    
+    print("   ✅ Created comprehensive AFCA dataset information")
+    return afca_info
+
+def save_metadata_files(metadata: Dict, afca_info: Dict, h5ad_info: Dict, dirs: Dict[str, Path]) -> None:
+    """Save all collected metadata to organized files."""
+    
+    print("\n💾 SAVING METADATA FILES")
+    print("=" * 50)
+    
+    try:
+        # Save GEO metadata
+        geo_file = dirs['metadata'] / 'geo_metadata.json'
+        with open(geo_file, 'w', encoding='utf-8') as f:
+            json.dump(metadata, f, indent=2, ensure_ascii=False)
+        print(f"   ✅ Saved GEO metadata: {geo_file}")
+        
+        # Save AFCA dataset information
+        afca_file = dirs['metadata'] / 'afca_dataset_info.json'
+        with open(afca_file, 'w', encoding='utf-8') as f:
+            json.dump(afca_info, f, indent=2, ensure_ascii=False)
+        print(f"   ✅ Saved AFCA info: {afca_file}")
+        
+        # Save h5ad processing results
+        h5ad_file = dirs['metadata'] / 'h5ad_processing_info.json'
+        with open(h5ad_file, 'w', encoding='utf-8') as f:
+            json.dump(h5ad_info, f, indent=2, ensure_ascii=False)
+        print(f"   ✅ Saved h5ad info: {h5ad_file}")
+        
+        # Create summary metadata
+        summary = {
+            'retrieval_date': pd.Timestamp.now().isoformat(),
+            'accession': metadata.get('accession', 'GSE218661'),
+            'title': metadata.get('title', afca_info['dataset_name']),
+            'organism': afca_info['data_description']['organism'],
+            'total_samples': metadata.get('sample_count', 'Unknown'),
+            'technology': afca_info['data_description']['technology'],
+            'h5ad_files_found': len(h5ad_info.get('files_found', [])),
+            'h5ad_files_processed': len(h5ad_info.get('files_processed', {})),
+            'total_cells_in_h5ad': h5ad_info.get('total_cells', 0),
+            'max_genes_in_h5ad': h5ad_info.get('total_genes', 0)
+        }
+        
+        summary_file = dirs['metadata'] / 'retrieval_summary.json'
+        with open(summary_file, 'w', encoding='utf-8') as f:
+            json.dump(summary, f, indent=2, ensure_ascii=False)
+        print(f"   ✅ Saved retrieval summary: {summary_file}")
+        
+        # Save sample information as CSV if available
+        if metadata.get('samples'):
+            samples_data = []
+            for gsm_id, sample_info in metadata['samples'].items():
+                row = {'sample_id': gsm_id}
+                row.update(sample_info)
+                # Flatten characteristics list
+                if isinstance(sample_info.get('characteristics_ch1'), list):
+                    for i, char in enumerate(sample_info['characteristics_ch1']):
+                        row[f'characteristic_{i+1}'] = char
+                samples_data.append(row)
+            
+            samples_df = pd.DataFrame(samples_data)
+            samples_file = dirs['metadata'] / 'samples_metadata.csv'
+            samples_df.to_csv(samples_file, index=False)
+            print(f"   ✅ Saved samples metadata: {samples_file}")
+        
+    except Exception as e:
+        print(f"   ❌ Error saving metadata: {e}")
+
+def generate_download_instructions() -> str:
+    """Generate instructions for manual data download."""
+    
+    instructions = """
+🔽 MANUAL DOWNLOAD INSTRUCTIONS FOR AFCA DATA
+==============================================
+
+1. GEO Repository (GSE218661) - Primary Source:
+   • Visit: https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE218661
+   • Download supplementary files (look for .h5ad files)
+   • Place in: data/raw/supplementary/
+
+2. AFCA Web Portal (Interactive):
+   • Visit: https://hongjielilab.org/afca/
+   • Access interactive data portal for exploration
+
+3. CellxGene Portal:
+   • Search for "Aging Fly Cell Atlas"
+   • URL: https://cellxgene.cziscience.com/
+
+4. Zenodo Repository (Analysis Code & Data):
+   • Visit: https://doi.org/10.5281/zenodo.7853649
+
+Expected h5ad files:
+- Head data: Contains head tissue single-nucleus data
+- Body data: Contains body tissue single-nucleus data  
+- Combined data: May contain integrated head+body data
+
+After manual download, place files in: data/raw/supplementary/
+Then re-run this script to process the downloaded files.
+"""
+    
+    return instructions
+
+def main():
+    """Main data retrieval workflow for AFCA GSE218661."""
+    
+    print("🧬 AGING FLY CELL ATLAS (AFCA) - DATA RETRIEVAL")
+    print("=" * 60)
+    print("🎯 Target: GSE218661 (Aging Fly Cell Atlas)")
+    print("📋 Goal: Download h5ad files and extract comprehensive metadata")
+    print()
+    
+    # Setup directories
+    dirs = setup_directories()
+    
+    # Extract GEO metadata and get GSE object
+    geo_metadata, gse = extract_geo_metadata("GSE218661")
+    
+    if not gse:
+        print("❌ Failed to retrieve GEO metadata. Cannot proceed.")
+        sys.exit(1)
+    
+    # Download supplementary files from GEO (this gets the sample directories but not h5ad files)
+    download_results = download_geo_supplementary_files(gse, dirs)
+    
+    # Manually download the h5ad files
+    h5ad_download_results = download_h5ad_files_manually(gse, dirs)
+    download_results.update(h5ad_download_results)
+    
+    # Process h5ad files
+    h5ad_info = process_h5ad_files(dirs)
+    
+    # Create comprehensive AFCA information
+    afca_info = create_afca_data_info()
+    
+    # Save all metadata
+    save_metadata_files(geo_metadata, afca_info, h5ad_info, dirs)
+    
+    # Generate download instructions
+    instructions = generate_download_instructions()
+    instructions_file = dirs['data'] / 'DOWNLOAD_INSTRUCTIONS.txt'
+    with open(instructions_file, 'w') as f:
+        f.write(instructions)
+    
+    # Final summary
+    print("\n🎉 DATA RETRIEVAL SUMMARY")
+    print("=" * 50)
+    
+    print("📂 Directory Structure Created:")
+    for name, path in dirs.items():
+        print(f"   ✅ {name}: {path}")
+    
+    print(f"\n📊 Download Results:")
+    for category, success in download_results.items():
+        status = "✅" if success else "❌"
+        print(f"   {status} {category}")
+    
+    print(f"\n🔬 H5AD Processing Results:")
+    print(f"   📁 Files found: {len(h5ad_info.get('files_found', []))}")
+    print(f"   ✅ Files processed: {len(h5ad_info.get('files_processed', {}))}")
+    if h5ad_info.get('total_cells', 0) > 0:
+        print(f"   🧬 Total cells: {h5ad_info['total_cells']:,}")
+        print(f"   🧮 Max genes: {h5ad_info['total_genes']:,}")
+    
+    print(f"\n📋 Metadata Files Created:")
+    metadata_files = [
+        'geo_metadata.json',
+        'afca_dataset_info.json', 
+        'h5ad_processing_info.json',
+        'retrieval_summary.json',
+        'samples_metadata.csv',
+        'DOWNLOAD_INSTRUCTIONS.txt'
+    ]
+    
+    for filename in metadata_files:
+        if filename == 'DOWNLOAD_INSTRUCTIONS.txt':
+            file_path = dirs['data'] / filename
+        else:
+            file_path = dirs['metadata'] / filename
+        if file_path.exists():
+            print(f"   ✅ {filename}")
+        else:
+            print(f"   ⚠️  {filename} (may not be created)")
+    
+    if not any([download_results.get('h5ad_head', False), download_results.get('h5ad_body', False)]):
+        print(f"\n⚠️  H5AD FILES NOT DOWNLOADED")
+        print("📖 Please check the manual download function or download directly from:")
+        print("🌐 Head: https://ftp.ncbi.nlm.nih.gov/geo/series/GSE218nnn/GSE218661/suppl/GSE218661_adata_head_S_v1.0.h5ad.gz")
+        print("🌐 Body: https://ftp.ncbi.nlm.nih.gov/geo/series/GSE218nnn/GSE218661/suppl/GSE218661_adata_body_S_v1.0.h5ad.gz")
+    else:
+        print(f"✅ Successfully downloaded h5ad files")
+    
+    if h5ad_info.get('files_processed'):
+        print(f"✅ Successfully processed {len(h5ad_info['files_processed'])} h5ad files")
+        
+        # Show tissue breakdown
+        tissues = {}
+        for filename, info in h5ad_info['files_processed'].items():
+            tissue = info.get('tissue_type', 'unknown')
+            if tissue not in tissues:
+                tissues[tissue] = {'files': 0, 'cells': 0}
+            tissues[tissue]['files'] += 1
+            tissues[tissue]['cells'] += info.get('n_obs', 0)
+        
+        print(f"\n📊 Tissue Breakdown:")
+        for tissue, stats in tissues.items():
+            print(f"   🧬 {tissue.title()}: {stats['files']} files, {stats['cells']:,} cells")
+    else:
+        print(f"⚠️  No h5ad files found or processed")
+        print("   Files may need to be downloaded manually or decompressed")
+    
+    print(f"\n🎯 NEXT STEPS:")
+    print("   1. Verify h5ad files were downloaded successfully")
+    print("   2. Check data/raw/supplementary/ for .h5ad.gz files")
+    print("   3. Run this script again to process downloaded files")
+    print("   4. Run 02_data_exploration.py to analyze the data")
+    print("   5. Visit AFCA web portal for interactive exploration")
+    
+    print(f"\n💾 All metadata saved to: {dirs['metadata']}")
+    print("🚀 Ready for data exploration phase!")
+
+if __name__ == "__main__":
+    main()