SeqKit - a cross-platform and ultrafast toolkit for FASTA/Q file manipulation

Introduction

FASTA and FASTQ are basic and ubiquitous formats for storing nucleotide and protein sequences. Common manipulations of FASTA/Q file include converting, searching, filtering, deduplication, splitting, shuffling, and sampling. Existing tools only implement some of these manipulations, and not particularly efficiently, and some are only available for certain operating systems. Furthermore, the complicated installation process of required packages and running environments can render these programs less user friendly.

This project describes a cross-platform ultrafast comprehensive toolkit for FASTA/Q processing. SeqKit provides executable binary files for all major operating systems, including Windows, Linux, and Mac OS X, and can be directly used without any dependencies or pre-configurations. SeqKit demonstrates competitive performance in execution time and memory usage compared to similar tools. The efficiency and usability of SeqKit enable researchers to rapidly accomplish common FASTA/Q file manipulations.

Table of Contents

Features

Features comparison

Categories Features seqkit fasta_utilities fastx_toolkit pyfaidx seqmagick seqtk
Formats support Multi-line FASTA Yes Yes -- Yes Yes Yes
FASTQ Yes Yes Yes -- Yes Yes
Multi-line FASTQ Yes Yes -- -- Yes Yes
Validating sequences Yes -- Yes Yes -- --
Supporting RNA Yes Yes -- -- Yes Yes
Functions Searching by motifs Yes Yes -- -- Yes --
Sampling Yes -- -- -- Yes Yes
Extracting sub-sequence Yes Yes -- Yes Yes Yes
Removing duplicates Yes -- -- -- Partly --
Splitting Yes Yes -- Partly -- --
Splitting by seq Yes -- Yes Yes -- --
Shuffling Yes -- -- -- -- --
Sorting Yes Yes -- -- Yes --
Locating motifs Yes -- -- -- -- --
Common sequences Yes -- -- -- -- --
Cleaning bases Yes Yes Yes Yes -- --
Transcription Yes Yes Yes Yes Yes Yes
Translation -- Yes Yes Yes Yes --
Filtering by size Yes Yes -- Yes Yes --
Renaming header Yes Yes -- -- Yes Yes
Other features Cross-platform Yes Partly Partly Yes Yes Yes
Reading STDIN Yes Yes Yes -- Yes Yes
Reading gzipped file Yes Yes -- -- Yes Yes
Writing gzip file Yes -- -- -- Yes --

Note 1: See version information of the softwares.

Note 2: See usage for detailed options of seqkit.

Subcommands

20 subcommands in total.

Sequence and subsequence

Format conversion

Searching

Set operations

Edit

Ordering

Misc

Installation

Go to Download Page for more download options and changelogs.

SeqKit is implemented in Go programming language, executable binary files for most popular operating systems are freely available in release page.

Method 1: Download binaries

Just download compressed executable file of your operating system, and decompress it with tar -zxvf *.tar.gz command or other tools. And then:

  1. For Linux-like systems

    1. If you have root privilege simply copy it to /usr/local/bin:

      sudo cp seqkit /usr/local/bin/
      
    2. Or add the current directory of the executable file to environment variable PATH:

      echo export PATH=\$PATH:\"$(pwd)\" >> ~/.bashrc
      source ~/.bashrc
      
  2. For windows, just copy seqkit.exe to C:\WINDOWS\system32.

Method 2: Install via conda Install-with-conda Anaconda Cloud downloads

conda install -c bioconda seqkit

Method 3: For Go developer

go get -u github.com/shenwei356/seqkit/seqkit

Technical details and guides for use

FASTA/Q format parsing

SeqKit uses author's lightweight and high-performance bioinformatics packages bio for FASTA/Q parsing, which has high performance close to the famous C lib klib (kseq.h).

Sequence formats and types

SeqKit seamlessly support FASTA and FASTQ format. Sequence format is automatically detected. All subcommands except for faidx can handle both formats. And only when some commands (subseq, split, sort and shuffle) which utilise FASTA index to improve perfrmance for large files in two pass mode (by flag --two-pass), only FASTA format is supported.

Sequence type (DNA/RNA/Protein) is automatically detected by leading subsequences of the first sequences in file or STDIN. The length of the leading subsequences is configurable by global flag --alphabet-guess-seq-length with default value of 10000. If length of the sequences is less than that, whole sequences will be checked.

Sequence ID

By default, most softwares, including seqkit, take the leading non-space letters as sequence identifier (ID). For example,

FASTA header ID
>123456 gene name 123456
>longname longname
>gi|110645304|ref|NC_002516.2| Pseudomona gi|110645304|ref|NC_002516.2|

But for some sequences from NCBI, e.g. >gi|110645304|ref|NC_002516.2| Pseudomona, the ID is NC_002516.2. In this case, we could set sequence ID parsing regular expression by global flag --id-regexp "\|([^\|]+)\| " or just use flag --id-ncbi. If you want the gi number, then use --id-regexp "^gi\|([^\|]+)\|".

FASTA index

For some commands, including subseq, split, sort and shuffle, when input files are (plain or gzipped) FASTA files, FASTA index would be optional used for rapid access of sequences and reducing memory occupation.

ATTENTION: the .seqkit.fai file created by SeqKit is slightly different from .fai file created by samtools. SeqKit uses full sequence head instead of just ID as key.

Parallelization of CPU intensive jobs

The validation of sequences bases and complement process of sequences are parallelized for large sequences.

Parsing of line-based files, including BED/GFF file and ID list file are also parallelized.

The Parallelization is implemented by multiple goroutines in golang which are similar to but much lighter weight than threads. The concurrency number is configurable with global flag -j or --threads (default value: 1 for single-CPU PC, 2 for others).

Memory occupation

Most of the subcommands do not read whole FASTA/Q records in to memory, including stat, fq2fa, fx2tab, tab2fx, grep, locate, replace, seq, sliding, subseq.

Note that when using subseq --gtf | --bed, if the GTF/BED files are too big, the memory usage will increase. You could use --chr to specify chromesomes and --feature to limit features.

Some subcommands need to store sequences or heads in memory, but there are strategy to reduce memory occupation, including rmdup and common. When comparing with sequences, MD5 digest could be used to replace sequence by flag -m (--md5).

Some subcommands could either read all records or read the files twice by flag -2 (--two-pass), including sample, split, shuffle and sort. They use FASTA index for rapid acccess of sequences and reducing memory occupation.

Reproducibility

Subcommands sample and shuffle use random function, random seed could be given by flag -s (--rand-seed). This makes sure that sampling result could be reproduced in different environments with same random seed.

Usage && Examples

Usage and examples

Tutorial

Benchmark

More details: http://bioinf.shenwei.me/seqkit/benchmark/

Datasets:

$ seqkit stat *.fa
file          format  type   num_seqs        sum_len  min_len       avg_len      max_len
dataset_A.fa  FASTA   DNA      67,748  2,807,643,808       56      41,442.5    5,976,145
dataset_B.fa  FASTA   DNA         194  3,099,750,718      970  15,978,096.5  248,956,422
dataset_C.fq  FASTQ   DNA   9,186,045    918,604,500      100           100          100

SeqKit version: v0.3.1.1

FASTA:

benchmark-5tests.tsv.png

FASTQ:

benchmark-5tests.tsv.png

Citation

W Shen, S Le, Y Li*, F Hu*. SeqKit: a cross-platform and ultrafast toolkit for FASTA/Q file manipulation. PLOS ONE. doi:10.1371/journal.pone.0163962.

Acknowledgements

We thank Lei Zhang for testing of SeqKit, and also thank Jim Hester, author of fasta_utilities, for advice on early performance improvements of for FASTA parsing and Brian Bushnell, author of BBMaps, for advice on naming SeqKit and adding accuracy evaluation in benchmarks. We also thank Nicholas C. Wu from the Scripps Research Institute, USA for commenting on the manuscript and Guangchuang Yu from State Key Laboratory of Emerging Infectious Diseases, The University of Hong Kong, HK for advice on the manuscript.

We thank Li Peng for reporting many bugs.

Contact

Email me for any problem when using seqkit. shenwei356(at)gmail.com

Create an issue to report bugs, propose new functions or ask for help.

License

MIT License