Skip to content

Latest commit

 

History

History
76 lines (62 loc) · 2.72 KB

kraken.md

File metadata and controls

76 lines (62 loc) · 2.72 KB
Raw

Kraken

Kraken operates by dissecting a given DNA sequence into smaller fragments known as k-mers. These k-mers are then cross-referenced with a comprehensive database. This database associates each k-mer with the Lowest Common Ancestor (LCA) of all genomes containing that specific k-mer. By doing this, Kraken can accurately predict the taxonomic classification of the original DNA sequence. We typically run Kraken on FASTQ reads to get an estimate of identification for each short read sequence, but it can also be used on FASTA files.

Pros

  • Visually intuitive
  • HTML files can be shared
  • Fast Analysis
  • Pre-built databases now available
  • Customizable

Cons

  • Large Database
  • High memory requirements
  • Not imune to database error and bias

Install

conda create -n kraken -c conda-forge -c bioconda kraken2 krona krakentools wget pandas pigz

After the conda install additional setup instructions are provided.

Download Kraken database.

There are many Databases to choose from. If unsure and download speeds allow try the standard database. If a smaller database is necessary Standard-8 may be a good option. Look at site for exact database naming.

Example download

cd ~; wget https://genome-idx.s3.amazonaws.com/kraken/k2_standard_08gb_20240605.tar.gz

mkdir k2_standard_08gb; tar -xzf k2_standard_08gb_*.tar.gz -C k2_standard_08gb

If needed link database to conda environment and download taxonomy.

rm -rf ${HOME}/miniconda3/envs/kraken/opt/krona/taxonomy
ln -s ${HOME}/k2_standard_08gb ${HOME}/miniconda3/envs/kraken/opt/krona/taxonomy
cd ~/k2_standard_08gb; ktUpdateTaxonomy.sh

Run sample

Kraken

db="${HOME}/k2_standard_08gb"
sample_name=$(echo *_R1*fastq.gz | sed 's/[_.].*//')
kraken2 --db ${db} --threads 4 --paired ERR766214_R1.fastq.gz ERR766214_R2.fastq.gz --output ${sample_name}-outputkraken.txt --report ${sample_name}-reportkraken.txt

Krona graph

kreport2krona.py --intermediate-ranks -r *reportkraken.txt -o ${sample_name}.krona
ktImportText ${sample_name}.krona -o ${sample_name}_krona.html
rm ${sample_name}.krona

Clean reads

This will extract all 533,222 reads of 77643 ID using SRR6856085 as example

taxid="77643"
name=`echo *_R1*.fastq.gz | head -1 | sed 's/[._].*//'`

(extract_kraken_reads.py \
    -k kraken/${name}_outputkraken.txt \
    -s1 ${name}_R1.fastq.gz \
    -s2 ${name}_R2.fastq.gz \
    -t ${taxid} \
    -o ${name}_tax${taxid}_R1.fastq \
    -o2 ${name}_tax${taxid}_R2.fastq \
    --fastq-output 2>&1) | grep -E "^(PROGRAM|>>|\t.*read IDs saved|\t.*reads printed|\tGenerated)" | grep -v "mill reads\|read IDs found\|million reads" | tee ${name}_tax${taxid}_output.log

HOME