example_hubmap.sh

#!/bin/bash
#
# Example script for downloading data from HuBMAP
# HuBMAP = Human BioMolecular Atlas Program
#
# This script demonstrates:
# - Downloading complete datasets using the /search endpoint
# - OR downloading limited results (~500-1000 items) using /param-search
#

set -e

echo "================================"
echo "HuBMAP Data Download Example"
echo "================================"
echo ""

# Create output directory
OUTPUT_DIR="./hubmap_data"
mkdir -p "$OUTPUT_DIR"

# Download the OpenAPI spec
echo "Step 1: Downloading HuBMAP Search API specification..."
SPEC_FILE="$OUTPUT_DIR/hubmap_search_spec.json"
curl -s "https://smart-api.info/api/metadata/7aaf02b838022d564da776b03f357158" > "$SPEC_FILE"
echo "✓ Spec downloaded to $SPEC_FILE"
echo ""

# Choose download method
echo "Step 2: Choose download method"
echo ""
echo "Option A: Complete datasets (recommended for research)"
echo "  Uses POST /search endpoint with Elasticsearch pagination"
echo "  Downloads ALL available data (may take longer)"
echo ""
echo "Option B: Quick sample (good for testing)"
echo "  Uses GET /param-search endpoint"
echo "  Downloads ~500-1000 items per entity type"
echo ""
read -p "Choose option (A/B) [B]: " OPTION
OPTION=${OPTION:-B}

if [ "$OPTION" = "A" ] || [ "$OPTION" = "a" ]; then
    echo ""
    echo "Step 3: Downloading COMPLETE datasets from HuBMAP..."
    echo "This will download ALL donors, samples, datasets, files, etc."
    echo "(This may take several minutes)"
    echo ""
    python3 openapi_downloader.py \
      --openapi "$SPEC_FILE" \
      --out-dir "$OUTPUT_DIR" \
      --use-search-api
else
    echo ""
    echo "Step 3: Downloading LIMITED sample data from HuBMAP..."
    echo "This will download up to ~500-1000 items per entity type"
    echo ""
    python3 openapi_downloader.py \
      --openapi "$SPEC_FILE" \
      --out-dir "$OUTPUT_DIR"
fi

echo ""
echo "================================"
echo "✓ Download complete!"
echo "================================"
echo ""
echo "CSV files saved to: $OUTPUT_DIR"
ls -lh "$OUTPUT_DIR"/*.csv 2>/dev/null || echo "No CSV files generated"
echo ""
echo "Example data in Donors.csv:"
head -n 3 "$OUTPUT_DIR/Donors.csv" 2>/dev/null || echo "Donors.csv not found"