Shovill: Robust Microbial Short-Read Genome Assembly

Torsten Seemann

doi:10.20944/preprints202607.1422.v1

Submitted:

18 July 2026

Posted:

20 July 2026

You are already at the latest version

Abstract

Shovill is a Unix command-line bioinformatics pipeline for performing de novo genome assembly of haploid microbial genomes from accurate paired-end short-read sequences, most commonly generated on the Illumina™ platform. Shovill is a full end-to-end workflow which accepts reads in FASTQ format and generates assembled contigs in FASTA format. It trims reads, sub-samples reads to an appropriate depth, corrects reads, stitches overlapping read pairs together,, de novo assembles the reads into contigs, and self-corrects any assembly errors. Each step is optimised to reduce RAM usage and maximise CPU usage. Shovill is written in Perl 5 and is available from github.com/tseemann/shovill under an open-source GPL 2.0 licence.

Keywords:

bioinformatics

;

microbiology

;

genomics

;

genome assembly

;

short reads

;

bacteria

;

software

Subject:

Biology and Life Sciences - Immunology and Microbiology

Impact Statement

Genome assembly is a fundamental data reduction step in microbial genomics. The public INSDC sequence repositories contain over 3 million microbial whole genome datasets, of which >90% are Illumina paired-end reads. Shovill has been optimised to efficiently and reliably assemble exactly this type of data, and was used successfully to assemble 2,763,297 Illumina genomes as part of the All-the-Bacteria project [1]. As of 2026, the software repository has been cited >1500 times[2], has been installed from Bioconda >60,000 times[3], has been run >160,000 times on the Galaxy servers in the last 5 years[4], and has >250 Github stars[5] placing it in the top 0.02% of software.

Data Summary

The source code for Shovill is available at https://github.com/tseemann/shovill under a GPL 2.0 open source licence, and is installable via the shovill package in Bioconda.

Introduction

Genome assembly is the process of reconstructing the DNA sequences of all the replicons in an organism, which for microbes, means all its chromosomes and plasmids. The vast majority of microbial genomes are sequenced using Illumina™ paired-end reads, which typically varied from 36 to 300 bp long, but today 150bp is most prevalent. Typically one aims for a sequencing yield that provides an average depth of 30 to 200 reads per genome position[6].

The earliest Illumina genome assembler software was Velvet[7], but the most popular assemblers for microbial isolates today are Spades[8] and SKESA[9]. These both accept input reads in FASTQ/FASTA format, and output contig sequences in FASTA format, and optionally an assembly graph in GFAformat[10]. From the beginning, assembly workflows were built that pre-processed the input reads to the main assembly tool, and post-processed the contigs to improve overall results.

Here we present Shovill, an open-source command-line software tool optimised specifically for de novo assembly of haploid microbial genomes from accurate paired-end short-read sequences. It performs a series of pre-processing and post-processing steps to improve the assembly result, maximise use of multiple CPU threads, reduce RAM and storage requirements, and increase the chance of successful completion of the assembly step. All parameters are estimated from the input data itself, and each workflow stage can disabled if desired.

Implementation

Input

Shovill requires only two Illumina paired-end FASTQ files, optionally gzip compressed, to be provided by the --R1 and --R2 options.

Genome Size Estimation

An estimate of the target genome size is needed by various downstream components of the Shovill pipeline. If not provided via the --gsize option, it will be estimated from the reads by using KMC[11] to count the number of unique 21-mers of frequency ≥ 10. This is a lower bound, as repeat regions within the genome will not generate unique k-mers.

Read Sub-Sampling

The sequencing yield in bacterial read sets is highly variable, ranging from 10x to 10,000x. For de novo genome assembly, more data is not better, because after a certain point we have saturated the signal and are only adding noise from sequencing errors[12]. The sweet spot for depth when assembling short reads is in the 50x to 200x range[6]. Shovill downsamples the input FASTQ files to 150x (or --depth) using SeqKit[13]. The sampling proportion is calculated as the quotient of the total FASTQ yield and the previously estimated genome size (or --gsize). This improves running time and makes resource requirements more predictable.

Read Trimming

If the --trim option is provided, Trimmomatic[14] is used to remove Illumina adaptors and low quality bases (≤ Q3) at read ends. The default however is to omit this step, because most Illumina instruments are configured to trim reads automatically, so not trimming reduces disk usage.

Read Error Correction

The RAM usage of a De Bruin graph based genome assembler grows linearly with the number of read sequences[15], so Shovill uses Lighter[16] to perform conservative read correction using options “-K 32 G -maxcor 1” where G is the genome size provided or estimated previously. This step can be disabled using the --noreadcorr option.

Read Stitching

Ideally, when performing Illumina paired-end sequencing, the DNA library is made up of fragments with lengths longer than twice the read length. If the library is over-fragmented, reads R1 and R2 will overlap and cause the “negative insert size” problem, which can lead to mis-assemblies or runtime software errors[17]. To prevent this, Shovill uses FLASH[18] to convert any overlapping pairs (minimum overlap 20 bp) into a single longer stitched read. Non-overlapping reads remain as separate paired reads. Stitching can be disabled using the --nostitch option.

Genome Assembly

Shovill does not prescribe specific software for the de novo assembly step. Instead, it allows the user to choose from four popular tools (Spades[8], SKESA[9],Velvet[7], or Megahit[19]) using the --assembler option. The most important parameter all of these assemblers need is a list of k-mer values to use. Although the user can provide an explicit list via --kmers, Shovill will automatically choose 5 evenly spaced k-mers in [K_min

-

K_max], where K_min=21 if the input reads are ≤ 75 bp long, or 31 otherwise; and K_max=127 or 75% of the input read length, whichever is smaller.

Genome Correction

At this point, we have our assembled contigs in FASTA format. Shovill now checks for consistency between the input reads and the output assembly. Firstly, the input reads are aligned to the assembly using BWA MEM[20] and stored in a sorted BAM using SAMtools[21]. Next, Pilon[22] uses the BAM to look for evidence of variants (small base differences, insertions or deletions), which are assumed to be minor misassemblies, and the contigs corrected. This step can be disabled with the --nocorr option.

Output

Shovill produces a final corrected FASTA file of assembled contigs. The contigs are sorted from longest to shortest, and the FASTA ID numbered using the --namefmt setting. The FASTA description is a series of name=value pairs which include the contig length (len). contig depth (cov), assembly date (date), and the assembler name and version details.

Discussion

Shovill has been optimised to efficiently and reliably assemble microbial genomes sequenced on the Illumina™ platform, the most common high-throughput sequence data type in the world, with over 3-million microbial samples uploaded to the INDSC repositories. In 2024 Shovill was adopted by the All-the-Bacteria project where it has been used to assemble 2,763,297 bacterial and archaeal genomes[1]. It can be run via the global online Galaxy plaform[23], or installed and run locally using Bioconda.

Author Contributions

T.S conceived, designed & implemented the software, and wrote the manuscript.

Funding Information

This research was partially supported by the Victorian Life Sciences Computation Initiative (VLSCI), an funding initiative of the Victorian State Government, Australia.

Ethical Statement

No sequence reads were harmed in this project.

Acknowledgments

The author would like to thank Anders Goncalves da Silva, Simon Gladman, Jason Kwong, Robert Petit III, Zamin Iqbal, Martin Hunt, John Lees, Jake Lacey, and Kristy Horan for their useful feedback.

Conflicts of Interest

The author declares that there are no conflicts of interest.

References

Hunt, M. et al.. AllTheBacteria: a community resource empowers biology and discovers novel peptide antibiotics. Bioinformatics (2024).
Citations to Shovill Github repository. Google Scholar https://scholar.google.com/scholar?q=github.com/tseemann/shovill.
Package Recipe ‘shovill’ — Bioconda documentation. https://bioconda.github.io/recipes/shovill/README.html.
Galaxy Codex: catalog of Galaxy resources (tools, training, workflows). Galaxy Communities Dock https://github.com/galaxyproject/galaxy_codex.
Seemann, T. Shovill: Assemble bacterial isolate genomes from Illumina paired-end reads. https://github.com/tseemann/shovill.
Desai, A. et al.. Identification of optimum sequencing depth especially for de novo genome assembly of small genomes using next generation sequencing data. PLoS One 8, e60204 (2013). [CrossRef] [PubMed]
Zerbino, D. R., McEwen, G. K., Margulies, E. H. & Birney, E. Pebble and rock band: heuristic resolution of repeats and scaffolding in the velvet short-read de novo assembler. PLoS One 4, e8407 (2009). [CrossRef] [PubMed]
Bankevich, A. et al.. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J. Comput. Biol. 19, 455–477 (2012). [CrossRef] [PubMed]
Souvorov, A., Agarwala, R. & Lipman, D. J. SKESA: strategic k-mer extension for scrupulous assemblies. Genome Biol. 19, 153 (2018). [CrossRef] [PubMed]
Dawson, E. T. & Durbin, R. GFAKluge: A C++ library and command line utilities for the Graphical Fragment Assembly formats. J. Open Source Softw. 4, 1083 (2019). [CrossRef] [PubMed]
Kokot, M., Dlugosz, M. & Deorowicz, S. KMC 3: counting and manipulating k-mer statistics. Bioinformatics 33, 2759–2761 (2017). [CrossRef] [PubMed]
Mirebrahim, H., Close, T. J. & Lonardi, S. De novo meta-assembly of ultra-deep sequencing data. Bioinformatics 31, i9–16 (2015). [CrossRef] [PubMed]
Shen, W., Sipos, B. & Zhao, L. SeqKit2: A Swiss army knife for sequence and alignment processing. Imeta 3, e191 (2024). [CrossRef] [PubMed]
Beier, S., Bolger, A. M., Bolger, M. E., Schwacke, R. & Usadel, B. Trimmomatic: A decade of feature-rich, high-performance NGS read preprocessing. Bioinformatics (2026). [CrossRef] [PubMed]
Chikhi, R. & Rizk, G. Space-efficient and exact de Bruijn graph representation based on a Bloom filter. Algorithms Mol. Biol. 8, 22 (2013). [CrossRef] [PubMed]
Song, L., Florea, L. & Langmead, B. Lighter: fast and memory-efficient sequencing error correction without counting. Genome Biol. 15, 509 (2014). [CrossRef] [PubMed]
Turner, F. S. Assessment of insert sizes and adapter content in fastq data from NexteraXT libraries. Front. Genet. 5, 5 (2014). [CrossRef] [PubMed]
Magoč, T. & Salzberg, S. L. FLASH: fast length adjustment of short reads to improve genome assemblies. Bioinformatics 27, 2957–2963 (2011). [CrossRef] [PubMed]
Li, D. et al.. MEGAHIT v1.0: A fast and scalable metagenome assembler driven by advanced methodologies and community practices. Methods 102, 3–11 (2016). [CrossRef] [PubMed]
Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. arXiv [q-bio.GN] (2013).
Li, H. et al.. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009). [CrossRef] [PubMed]
Walker, B. J. et al.. Pilon: an integrated tool for comprehensive microbial variant detection and genome assembly improvement. PLoS One 9, e112963 (2014). [CrossRef] [PubMed]
Galaxy Community. The Galaxy platform for accessible, reproducible, and collaborative data analyses: 2024 update. Nucleic Acids Res. 52, W83–W94 (2024). [CrossRef] [PubMed]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2026 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.