Start Silva Server VM

Overview

This script starts a taxonomic server for Silva ribosomal RNA identification using BBTools' taxserver.sh. The server provides a web-based interface for identifying ribosomal RNA sequences through k-mer sketching against the Silva database. It's specifically configured to run on JGI's jgi-web-4 server infrastructure.

Warning: This script is designed specifically for JGI infrastructure and contains hardcoded paths and server configurations. It should not be used outside the intended environment without modification.

Server Configuration

Service Settings

LOG: Log file: ribologVM_32.txt - Contains server output and error messages
PASS: Authentication password: xxxxx (placeholder - real password configured in production)
DOMAIN: Service domain: https://ribo-sketch.jgi.doe.gov - Public endpoint for the service
KILL: Kill URL: https://ribo-sketch.jgi.doe.gov/kill/ - Administrative endpoint for server termination
PORT: Service port: 3073 - Internal port for the taxonomic server
REF: Reference sketches: /global/projectb/sandbox/gaag/bbtools/silva/latest/both_seq#.sketch - Silva database sketches
DB: Database name: silva - Identifier for the Silva ribosomal RNA database

Server Parameters

Taxserver Configuration

-da: JVM assertion flag - Disables assertions for production performance
-Xmx10g: Java heap memory: 10GB - Maximum memory allocation for the server process
port=$PORT: Server port: Uses PORT variable (3073) for incoming connections
verbose: Verbose logging: Enables detailed output for debugging and monitoring
tree=auto: Taxonomic tree: Automatic tree loading and construction
sketchonly: Sketch-only mode: Uses only k-mer sketches for identification (faster)
index: Index loading: Loads reference sketches into memory for fast queries
whitelist: Access control: Enables IP whitelisting for security
domain=$DOMAIN: Web domain: Sets the public domain for web interface access
killcode=$PASS: Kill authentication: Password required for server termination
oldcode=$PASS: Legacy authentication: Backward compatibility authentication code
oldaddress=$KILL: Legacy kill URL: Backward compatibility termination endpoint
ref=$REF: Reference database: Path to Silva sketch files for sequence matching
dbname=Silva: Database display name: Human-readable name shown in web interface
blacklist=silva: Blacklist configuration: Silva-specific filtering rules
k=32,24: K-mer sizes: Uses both 32-mer and 24-mer sketches for identification accuracy

Prerequisites

System Requirements

Server: JGI jgi-web-4 infrastructure
Java: Java 8 or later with 10GB+ heap space
BBTools: Complete BBTools installation with taxserver.sh
Memory: Minimum 12GB RAM (10GB for Java heap + system overhead)
Storage: Access to Silva sketch database files

Database Requirements

Silva Sketches: Pre-computed k-mer sketches of Silva ribosomal RNA sequences
Path Access: Read access to /global/projectb/sandbox/gaag/bbtools/silva/latest/
File Pattern: Sketch files matching both_seq*.sketch pattern
Database Version: Latest Silva release with taxonomic annotations

Network Requirements

Port Access: Port 3073 must be available for binding
Domain Registration: DNS entry for ribo-sketch.jgi.doe.gov
SSL Certificate: Valid certificate for HTTPS service
Firewall Rules: Appropriate access controls for public service

Usage

Starting the Server

Run the script directly on the designated server:

./startSilvaServerVM.sh

The server will start in the background using nohup and log output to ribologVM_32.txt.

Testing Mode

For development and testing, the script includes a simplified command (commented out):

# Simple mode for testing
/global/projectb/sandbox/gaag/bbtools/jgi-bbtools/taxserver.sh -ea -Xmx10g \
    port=3073 verbose tree=auto sketchonly silva k=32,24 index=f \
    domain=https://ribo-sketch.jgi-psf.org

This testing mode uses:

-ea: Enables assertions for debugging
index=f: Disables index loading for faster startup
Alternative domain: jgi-psf.org instead of jgi.doe.gov
No authentication: Simplified configuration for testing

Service Management

Monitoring

Log File: Monitor ribologVM_32.txt for server status and errors
Process Status: Use ps aux | grep taxserver to check running status
Port Status: Use netstat -tlnp | grep 3073 to verify port binding
Web Interface: Access https://ribo-sketch.jgi.doe.gov for service availability

Stopping the Server

Administrative Stop: Use the kill URL with authentication password
Process Kill: Find and kill the Java process running taxserver.sh
Graceful Shutdown: Use standard SIGTERM signal for clean shutdown

Troubleshooting

Port Conflicts: Ensure port 3073 is not in use by other services
Memory Issues: Monitor system memory usage; increase -Xmx if needed
Database Access: Verify read permissions for Silva sketch files
Network Connectivity: Test domain resolution and SSL certificate validity
Authentication: Verify password configuration for kill functionality

Security Considerations

Access Control

Whitelist: Server implements IP-based access control
Authentication: Kill functionality requires password authentication
HTTPS: All communication encrypted via SSL/TLS
Process Isolation: Server runs as dedicated service user

Data Protection

Read-Only Database: Silva sketches accessed in read-only mode
No User Data Storage: Query sequences processed in memory only
Log Rotation: Regular rotation of log files to prevent disk filling
Resource Limits: JVM heap size limits prevent memory exhaustion

Service Architecture

Component Overview

The Silva server provides ribosomal RNA identification through the following components:

Web Interface: HTTPS endpoint for sequence submission and results
Sketch Engine: K-mer sketching for fast sequence comparison
Taxonomic Database: Silva ribosomal RNA sequences with taxonomic annotations
Query Processing: Real-time sequence analysis and classification
Result Formatting: Structured output with taxonomic assignments

Performance Characteristics

Query Speed: Sub-second response for typical rRNA sequences
Throughput: Multiple concurrent queries supported
Memory Usage: 10GB heap for sketch indexing and query processing
Accuracy: Dual k-mer strategy (32,24) for robust identification
Database Coverage: Comprehensive Silva ribosomal RNA collection

Integration

API Access

The server provides programmatic access through HTTP/HTTPS requests to the configured domain. Queries can be submitted as FASTA sequences for taxonomic identification against the Silva database.

Related Tools

taxserver.sh: Core taxonomic server implementation
sketch.sh: Manual sketching tool for sequence analysis
sendsketch.sh: Client tool for submitting queries to taxonomic servers
comparesketch.sh: Local sketch comparison and analysis

Database Updates

Silva database sketches should be updated periodically to reflect new Silva releases. The server should be restarted after sketch updates to load the latest data.

Notes

This script is specifically designed for JGI's production infrastructure
Hardcoded paths and configurations require modification for other environments
The service provides public access to Silva-based ribosomal RNA identification
Regular monitoring of log files and system resources is recommended
Authentication passwords should be properly secured in production deployments