# Install BLAST+
conda install -c bioconda blast

# Or via package manager
# macOS: brew install blast
# Ubuntu: sudo apt-get install ncbi-blast+

# Install MCP server
git clone https://github.com/bio-mcp/bio-mcp-blast.git
cd bio-mcp-blast
pip install -e .

# Start the server
python -m src.server

# Or with queue support
python -m src.main --mode queue

{
  "mcpServers": {
    "bio-blast": {
      "command": "python",
      "args": ["-m", "src.server"],
      "cwd": "/path/to/bio-mcp-blast"
    }
  }
}

User: "BLAST this sequence against nr: ATGCGATCGATCG"
AI: [calls blastn] → Returns top hits with E-values and alignments

User: "Search proteins.fasta against SwissProt database"
AI: [calls blastp] → Processes file and returns similarity results

User: "Create a BLAST database from reference_genomes.fasta"
AI: [calls makeblastdb] → Creates searchable database files

User: "BLAST large_dataset.fasta against nt database"
AI: [calls blastn_async] → "Job submitted! ID: abc123, checking progress..."

# Basic settings
export BIO_MCP_MAX_FILE_SIZE=100000000    # 100MB max file size
export BIO_MCP_TIMEOUT=300                # 5 minute timeout
export BIO_MCP_BLAST_PATH="blastn"        # BLAST executable path

# Queue mode settings
export BIO_MCP_QUEUE_URL="http://localhost:8000"

# Download common databases
mkdir -p ~/blast-databases
cd ~/blast-databases

# NCBI databases (large downloads!)
update_blastdb.pl --decompress nt
update_blastdb.pl --decompress nr
update_blastdb.pl --decompress swissprot

# Set environment variable
export BLASTDB=~/blast-databases

# Build image
docker build -t bio-mcp-blast .

# Run container
docker run -p 5000:5000 \
  -v ~/blast-databases:/data/blast-db:ro \
  -e BLASTDB=/data/blast-db \
  bio-mcp-blast

services:
  blast-server:
    build: .
    ports:
      - "5000:5000"
    volumes:
      - ./databases:/data/blast-db:ro
    environment:
      - BLASTDB=/data/blast-db
      - BIO_MCP_TIMEOUT=600

# Start queue infrastructure
cd ../bio-mcp-queue
./setup-local.sh

# Start BLAST server with queue support
python -m src.main --mode queue --queue-url http://localhost:8000

# Submit async job
job_info = await blast_server.submit_job(
    job_type="blastn",
    parameters={
        "query": "large_sequences.fasta",
        "database": "nt",
        "evalue": 0.001
    }
)

# Check status
status = await blast_server.get_job_status(job_info["job_id"])

# Get results when complete
results = await blast_server.get_job_result(job_info["job_id"])

# Fields: query_id, subject_id, percent_identity, alignment_length, ...
Query_1    gi|123456    98.5    500    7    0    1    500    1000    1499    1e-180    633

{
  "BlastOutput2": [{
    "report": {
      "results": {
        "search": {
          "query_title": "Query_1",
          "hits": [...]
        }
      }
    }
  }]
}

# Run tests
pytest tests/ -v

# Test with real data
python tests/test_integration.py

# Performance testing
python tests/benchmark.py

# Check installation
which blastn
blastn -version

# Install via conda
conda install -c bioconda blast

# Check BLASTDB environment variable
echo $BLASTDB

# List available databases
blastdbcmd -list /path/to/databases

# Reduce max_target_seqs
blastn -max_target_seqs 100

# Use streaming for large outputs
# Increase system swap space

# Increase timeout
export BIO_MCP_TIMEOUT=3600  # 1 hour

# Or use queue mode for long searches
python -m src.main --mode queue