Kraken2 Classification - Usage Guide

Kraken2 Classification - Usage Guide

Overview

Kraken2 is a fast taxonomic classifier that uses exact k-mer matches to assign reads to taxonomic nodes. It's highly accurate for well-represented taxa and ideal for screening large datasets.

Prerequisites

conda install -c bioconda kraken2

# Download pre-built database
wget https://genome-idx.s3.amazonaws.com/kraken/k2_standard_08gb_20230605.tar.gz
mkdir kraken2_db && tar -xzf k2_standard_08gb_20230605.tar.gz -C kraken2_db

# Or build standard database (~100GB disk, ~50GB final)
kraken2-build --standard --db kraken2_standard_db --threads 16

Quick Start

Tell your AI agent what you want to do:

• "Classify my metagenomic reads to identify organisms"
• "Run Kraken2 on paired-end reads and generate a report"
• "Filter out human reads from my metagenome"

Example Prompts

Basic Classification

"Run Kraken2 on reads_R1.fastq.gz and reads_R2.fastq.gz using the standard database"

"Classify my shotgun metagenomics data and show me the top 20 species"

Database Selection

"Use the viral database to screen for viral sequences in my sample"

"Run Kraken2 with the standard-8 database since I have limited RAM"

Post-classification Analysis

"Extract only bacterial classifications from my Kraken2 output"

"Get species-level results and sort by abundance"

What the Agent Will Do

1. Verify Kraken2 installation and database availability
2. Run classification with appropriate parameters for paired/single reads
3. Generate per-read classifications and taxonomic report
4. Parse and summarize results by taxonomic level

Tips

• Use --memory-mapping for low-memory systems (slower but works)
• Classification rate of 30-70% is normal for environmental samples
• Store databases on SSD for better performance
• Run Bracken after Kraken2 for improved abundance estimates
• Use --threads to parallelize (8-16 typically optimal)

Output Files

Resources

• Kraken2 GitHub
• Pre-built Databases