The bash script uses TaxonKit to produce a list of taxons to filter, Seqkit to extract the reads, and csvtk for data wrangling. Written for my own purposes.
Use conda to install these:
conda install -c bioconda seqkit taxonkit csvtk
Usage: extract_reads.sh -f sample.r1.fastq.gz -r sample.r2.fastq.gz -t 10240 -k sample.kraken.txt
-f | --forward_fastq : R1 fastq.gz file
-r | --reverse_fastq : R2 fastq.gz file
-k | --kraken : Kraken output of classified reads
-t | --taxon : NCBI taxon ID
-j | --threads : Number of threads to use
-h | --help : This message
Taxonkit is used to produce a subtree of given TaxIds. For example, given taxonid 10244, this script will extract all reads with the following taxon IDs:
% taxonkit list --ids 10245 --indent '' | tr '\n' ' '
10245 10246 10247 10248 10249 10250 10251 10252 10253 10254
31531 32605 32606 45417 124313 126794 130665 130666 201444
262397 301352 332193 350831 350832 397342 467144 502057 691321
696871 1651169
extract_kraken_reads.py is commonly used to extract reads classified by kraken. Unfortunately for large libraries it can be slow.
This script is designed with speed in mind and not as feature rich as extract_kraken_reads.py.
- Allow user to provide a list of taxons, skipping the need for
taxonkit - Allow user to specify flags
--childrenor--ancestor