BUSCO applications from quality assessments to gene prediction and phylogenomics
Robert Michael WaterhouseiD, Mathieu SeppeyiD, Felipe A. SimãoiD, Mosè ManniiD, Panagiotis IoannidisiD, Guennadi Klioutchnikov, Evgenia V. KriventsevaiD, Evgeny M. ZdobnoviD
Abstract
Genomics promises comprehensive surveying of genomes and metagenomes, but rapidly changing technologies and expanding data volumes make evaluation of completeness a challenging task. Technical sequencing quality metrics can be complemented by quantifying completeness in terms of the expected gene content of Benchmarking Universal Single-Copy Orthologs (BUSCO, http://busco.ezlab.org). Now in its third release, BUSCO utilities extend beyond quality control to applications in comparative genomics, gene predictor training, metagenomics, and phylogenomics.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 2,214
- FWCI
- 87.86
- Field percentile
- top 1%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 1 (17 Aug 2017) · CC BY 4.0
BUSCO v3 quantifies genome, gene set and transcriptome completeness via near-universal single-copy orthologs and repurposes them for gene predictor training, reference selection, and phylogenomic marker discovery.
Fast-changing sequencing technologies and growing data volumes make the completeness of assemblies, gene sets and transcriptomes hard to judge, and the widely used Core Eukaryotic Genes Mapping Approach (CEGMA) has been discontinued. BUSCO measures completeness against expected single-copy ortholog content for like-for-like comparisons, and its third release adds refactored code, configuration-file control, plotting, and uses beyond quality control.
Key results
ShownSuccessive releases of the honeybee and chicken assemblies and annotated gene sets show stepwise completeness gains, demonstrating that BUSCO can track the effect of re-assembly and re-annotation efforts.
Completeness rises from initial to intermediate to latest versions for both species under the Hymenoptera or Aves lineage sets and under the Metazoa set (Figure 1); per-version percentages are not stated in the text. · Abstract; Figure legends
ShownAugustus parameters trained on BUSCO-identified complete genes improve ab initio gene prediction where no pre-trained parameters exist, and roughly match or occasionally exceed species-specific trained parameter sets.
Strigamia centipede, Daphnia waterflea and Danaus butterfly predictions beat those using fruit fly parameters; tomato and thale cress almost equal, fruit fly and Nasonia equal, Tribolium better than species-specific sets; scored as percent sequence length match to official gene sets (Figure 2, Figure S1). · Abstract; Figure legends
ShownBUSCO runs on rodent genomes and transcriptomes recover shared complete single-copy orthologs whose superalignment gives a maximum likelihood species tree consistent with earlier studies, with topology robust to the lineage dataset used.
Seven rodent genomes and five transcriptomes; Euarchontoglires-based tree rooted with rabbit had 100% bootstrap support at all nodes; Mammalia and Metazoa gave identical topologies; Euarchontoglires or Mammalia found more than three times as many markers as Metazoa but ran longer. · Abstract; Figure legends
SuggestedCompleteness scores offer objective criteria for choosing comparative genomics representatives; gene count is a poor proxy, and RefSeq-designated reference genomes are not always the most complete available.
Only 63% of 653 Streptomyces genomes held the full set of complete bacteria BUSCOs; assessing 135 Lactobacillus and 35 Aspergillus genomes against contiguity and total gene counts shows RefSeq references are sometimes outperformed (Figure S2, not reproduced in text). · Abstract
ClaimedBUSCO completeness estimates broadly agree with independent completeness measures in vertebrates and plants.
Alignment-based completeness of ultraconserved elements and coding exons for 20 vertebrates against human agreed very well with BUSCO; a 12-plant comparison with CEGMA, core plant gene families and EST mapping also agreed; no concordance statistics are given. · Abstract
Methods, data and reproducibility
BUSCO searches genome assemblies, gene sets or transcriptomes with sequence profiles of lineage-specific near-universal single-copy orthologs drawn from OrthoDB v9, labeling each as complete (single-copy or duplicated), fragmented or missing; genome assessments run Augustus, yielding trained parameters and gene models reusable downstream.
- Honeybee genome assemblies and annotated gene sets, initial, intermediate and latest versions · honeybee · n = three versions
- Chicken genome assemblies and annotated gene sets, initial, intermediate and latest versions · chicken · n = three versions
- Genomes used to test BUSCO-trained Augustus (centipede, waterflea, Danaus butterfly, fruit fly, Nasonia, Tribolium, tomato, thale cress) · multiple species (arthropods and plants)
- Streptomyces genomes surveyed for complete bacteria BUSCOs · Streptomyces (bacteria) · n = 653
- Lactobacillus genomes assessed alongside contiguity and total gene counts · Lactobacillus (bacteria) · n = 135
- Aspergillus genomes assessed alongside contiguity and total gene counts · Aspergillus (fungi) · n = 35
- Rodent genomes and transcriptomes for phylogenomic marker recovery; tree rooted with rabbit · rodents; rabbit as root · n = seven genomes and five transcriptomes
- Alternative completeness comparisons: vertebrate genomes aligned to human; plant genomes assessed with several methods · vertebrates (including human); plants · n = 20 vertebrates; 12 plants
- Methods & tools
- BUSCO v3, BUSCO v2, BUSCO plotting tool (bar charts), Augustus (ab initio gene prediction), SNAP (gene predictor trainable on BUSCO GFF/GenBank models), OrthoDB v9 (ortholog source for lineage datasets), BUSCO lineage datasets: Metazoa, Hymenoptera, Aves, Mammalia, Euarchontoglires, bacteria, CEGMA (Core Eukaryotic Genes Mapping Approach), RefSeq reference genome designations, Maximum likelihood tree reconstruction with bootstrap support, Core plant Gene Families and Expressed Sequence Tag (EST) mapping (comparators)
- Data & accessions
- http://www.orthodb.org/
- Software & environment
- BUSCO v3: refactored code with configuration-file control and a bar-chart plotting tool; distributed via GitLab, as an Ubuntu virtual machine, and as an online service for logged-in OrthoDB users. Lineage datasets built from OrthoDB v9.
- Reproducibility gaps
- Methods, data availability statements and accession codes are only in supplementary material absent from the text; the GitLab repository URL is not printed, and the OrthoDB online service requires login.
Limitations (5)
- Because gene prediction and orthology assignment are error-prone, BUSCO classifications can fall short of 100% correctness even with score and length cut-offs tuned for precision and recall. · stated by the authors
- Lineage datasets must be built from currently sequenced species, which are not evenly spread across the phylogeny, even though closely related species are deliberately not over-sampled. · stated by the authors
- Completeness is only one input to reference choice; taxonomic sampling, available functional genomics data, annotation extent or accuracy, and historical usage also shape which genomes should be selected. · stated by the authors
- Gene-prediction, quality-control and reference-selection comparisons appear only graphically or qualitatively in the main text, with methods and accessions deferred to supplementary material, so effect sizes cannot be checked here. · noted in this summary
- Augustus accuracy is judged against official gene sets treated as ground truth, which may themselves contain errors, and agreement with alternative completeness metrics is reported without quantitative concordance measures. · noted in this summary
Who should read:Read this if you assess assembly, annotation or transcriptome completeness, need training data for ab initio gene predictors, or want single-copy ortholog markers for phylogenomics or reference genome selection.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Assessing species coverage and assembly quality of rapidly accumulating sequenced genomes
Romain Feron, Robert Michael Waterhouse·GigaSciencePeer-reviewed·Jan 1, 2022
- Universal orthologs infer deep phylogenies and improve genome quality assessments
Md Nafis Ul Alam, Cristian Román‐Palacios, … Rod A. Wing·BMC BiologyPeer-reviewed·Jul 27, 2025
- miniBUSCO: a faster and more accurate reimplementation of BUSCO
Neng Huang, Heng Li·bioRxiv·Jun 3, 2023
- nf-core/genomeqc: a best-practice pipeline for comparing genome and assembly quality
Christopher D. R. Wyatt, Fernando Duarte Frutos, … Seirian Sumner·bioRxiv·Aug 20, 2026
- Relating genome completeness to functional predictions
Jessica Liu, Tre’Andice Williams, John A. Burns·bioRxiv·Oct 1, 2021
- CoCoPyE: feature engineering for learning and prediction of genome quality indices
Niklas Birth, Nicolina Leppich, … Peter Meinicke·GigaSciencePeer-reviewed·Jan 1, 2024