ncbi::sra-human-scrubber

Description

The human read removal tool (HRRT) is based on the SRA Taxonomy Analysis Tool that will take as input a fastq file, and produce as output a fastq.clean file in which all reads identified as potentially of human origin are masked with ‘N’.

Overview

Briefly, the HRRT employs a k-mer database constructed of k-mers from Eukaryota derived from all human RefSeq records and subtracts any k-mers found in non-Eukaryota RefSeq records. The remaining set of k-mers compose the database used to identify human reads by the removal tool. This means the tool tends to be aggressive about identifying human reads since it contains not only human-specific k-mers, but also k-mers common to primates and other taxa further up the Eukaryotic tree. However, it is also fairly conservative at retaining any viral or bacterial clinical pathogen sequences. It takes a fastq file as input, identifies any reads with hits to the ‘human’ k-mer database and outputs a fastq.clean with the identified human reads masked with ‘N’.

Quick Start

• Clone the repo.
• pushd or cd to directory sra-human-scrubber.
  • Alternatively, download the zip file from the green ‘Code’ button, unzip it, then cd to directory sra-human-scrubber-master.
• Execute ./init_db.sh in directory sra-human-scrubber - this will retrieve the default (newest) pre-built db from ftp and place it in the directory sra-human-scrubber/data where it needs to be located.
• Please note binary aligns_toin bin was compiled on x86_64 GNU/Linux.
• Please refer to CHANGELOG for recent changes.

Usage

Working in the directory sra-human-scrubber (or alternatively sra-human-scrubber-master)

Invoke the test

Here the command is simply given the (file) argument test ./scripts/scrub.sh test

./scripts/scrub.sh test
2022-08-31 14:35:08    aligns_to version 0.707
2022-08-31 14:35:08    hardware threads: 32, omp threads: 32
2022-08-31 14:35:09    loading time (sec) 1
2022-08-31 14:35:09    /tmp/tmp.EpHdBbPYzb/temp.fasta
2022-08-31 14:35:09    FastaReader
2022-08-31 14:35:09    100% processed
2022-08-31 14:35:09    total spot count: 2
2022-08-31 14:35:09    total read count: 2
2022-08-31 14:35:09    total time (sec) 1
1  spot(s) masked or removed.

test succeeded

Mask human reads from fastq file

Here the command is given the path to your local fastq file as argument ./scripts/scrub.sh path-to-fastq-file/filename.fastq

Example: ./scripts/scrub.sh $TmpRuns/MyFastqFile.fastq

./scripts/scrub.sh $TmpRuns/MyFastqFile.fastq 
2022-08-31 14:43:58    aligns_to version 0.707
2022-08-31 14:43:58    hardware threads: 32, omp threads: 32
2022-08-31 14:43:59    loading time (sec) 1
2022-08-31 14:43:59    /tmp/tmp.jZwaayxNAA/temp.fasta
2022-08-31 14:43:59    FastaReader
2022-08-31 14:43:59    0% processed
2022-08-31 14:44:00    100% processed
2022-08-31 14:44:00    total spot count: 216859
2022-08-31 14:44:00    total read count: 216859
2022-08-31 14:44:00    total time (sec) 2
129  spot(s) masked or removed.

ls -l $TmpRuns/
-rw-r--r-- 1   78656910 Aug 31 14:43 MyFastqFile.fastq
-rw-r--r-- 1   78656910 Aug 31 14:44 MyFastqFile.fastq.clean

Note by default the application scales to use all threads available ( see option -p for setting threads below ).

Docker container available here: https://hub.docker.com/r/ncbi/sra-human-scrubber

Other useful options:

./scripts/scrub.sh -h
Usage: scrub.sh [OPTIONS] [file.fastq] 
OPTIONS:
    -i <input_file>; Input Fastq File.
    -o <output_file>; Save cleaned sequence reads to file, or set to - for stdout.
        NOTE: When stdin is used, output is stdout by default.
    -p <number> Number of threads to use.
    -d <database_path>; Specify a database other than default to use.
    -x ; Remove spots instead of default 'N' replacement.
        NOTE: Now by default sequence length of identified spots replaced with 'N'.
    -r ; Save identified spots to <input_file>.spots_removed.
    -u <user_named_file>; Save identified spots to <user_named_file>.
        NOTE: Required with -r if output is stdout, otherwise optional.
    -t ; Run test.
    -s ; Input is (collated) interleaved paired-end(read) file AND you wish both reads masked or removed.
    -h ; Display this message.

Note on Additional Testing

Internally the core scrubber binary (aligns_to) is subject to a Constant Integration (CI) regimen employing automatic testing with any code change. Using two SRA records with significant amounts of human reads, we test that the expected human reads are identified and in one case that the small amount of SARS-CoV-2 reads falsely identified as human is limited to those currently expected.