BlobToolKit Interactive quality assessment of genome assemblies
RICHARD J. CHALLISiD, Edward RichardsiD, Jeena RajaniD, Guy CochraneiD, Mark BlaxteriD
Abstract
Reconstruction of target genomes from sequence data produced by instruments that are agnostic as to the species-of-origin may be confounded by contaminant DNA. Whether introduced during sample processing or through co-extraction alongside the target DNA, if insufficient care is taken during the assembly process, the final assembled genome may be a mixture of data from several species. Such assemblies can confound sequence-based biological inference and, when deposited in public databases, may be included in downstream analyses by users unaware of underlying problems. We present BlobToolKit, a software suite to aid researchers in identifying and isolating non-target data in draft and publicly available genome assemblies. BlobToolKit can be used to process assembly, read and analysis files for fully reproducible interactive exploration in the browser-based Viewer. BlobToolKit can be used during assembly to filter non-target DNA, helping researchers produce assemblies with high biological credibility. We have been running an automated BlobToolKit pipeline on eukaryotic assemblies publicly available in the International Nucleotide Sequence Data Collaboration and are making the results available through a public instance of the Viewer at https://blobtoolkit.genomehubs.org/view. We aim to complete analysis of all publicly available genomes and then maintain currency with the flow of new genomes. We have worked to embed these views into the presentation of genome assemblies at the European Nucleotide Archive, providing an indication of assembly quality alongside the public record with links out to allow full exploration in the Viewer.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 2,650
- FWCI
- 58.69
- Field percentile
- top 1%
- Organism
- Assay
- Research Focus
- Approach
In-depth summary
Claude Opus 5.5 · xhigh effort · 8 Nov 2026 · from the full text of version 1 (15 Nov 2019) · CC BY 4.0
BlobToolKit combines GC proportion, read coverage and taxonomic hits per contig in an interactive browser Viewer, exposing contaminants and cobionts across public eukaryotic genome assemblies.
Contaminant or symbiont DNA sequenced alongside a target species can be assembled into its draft genome and, once deposited publicly, misattribute genes or traits to the wrong organism in downstream analyses. BlobToolKit reimplements BlobTools with an extensible, validated JSON format, real-time browser exploration, URL-encoded reproducible filters and an automated pipeline that screens INSDC eukaryotic assemblies and feeds ENA pages.
Key results
ShownThe thicket tinamou assembly PTEZ01 contains a low-coverage, GC-rich scaffold blob attributable to a Sarcocystis-like apicomplexan; parasite-derived sequence also occurs in bobwhite quail and sperm whale assemblies, including a RefSeq representative genome.
PTEZ01 (1.1 Gb): ~130 Mb of scaffolds at order-of-magnitude lower coverage, mean GC 0.52 vs 0.42 for the main assembly; 58 Mb assigned to Eucoccidiorida (Sarcocystis matches), 24 Mb to sperm whale and 8 Mb to bobwhite; bobwhite AWGT02 has 48 Mb of 1.2Gb (4%) apicomplexan; sperm whale PGGR02 holds 4.3 Mb assigned to Apicomplexa. · Identification of mis-annotated records in…
ShownDrosophila albomicans assembly ACVV01 contains proteobacterial cobiont genomes (Acetobacter, Gluconobacter) separable by higher GC and lower coverage, whereas Wolbachia-assigned scaffolds track host GC and coverage, consistent with nuclear insertions rather than a symbiont genome.
1,440 scaffolds best match Proteobacteria (nt and UniProt); of 6.7 Mb proteobacterial span, 4.3 Mb is Acetobacter and 1.8 Mb Gluconobacter (typical genomes 3-4 Mb); Wolbachia-assigned scaffolds total only 190 kb versus ~1.4 Mb for insect-infecting strains, with host-like GC and coverage. · Identification of common cobionts; Figure legends
ShownCobiont scaffolds can carry BUSCO reference genes, so completeness scores and phylogenomic datasets may silently absorb contaminant loci; misplaced dipteran BUSCOs sit on bacterial scaffolds in one Drosophila assembly and on arthropod scaffolds in another, and vice versa.
In ACVV01, several diptera_odb9 BUSCOs map to Proteobacteria-assigned scaffolds, five of which show bacterial-like GC and coverage; excluding them shifts completeness only slightly (83.7% vs 83.9%; BUSCO 3.0.2); in D. elegans AFFF02 two such BUSCOs sit on Proteobacteria scaffolds; PTEZ01 putative Sarcocystis scaffolds give 6% of unique eukaryote BUSCO hits. · Identification of common cobionts; Identification of mis-annotated records in…
ShownBinned plots and default no-hit filtering keep the highly fragmented cone snail assembly SDAX01 browsable; cross-library coverage plots flag questionable low-coverage scaffolds, and a kite plot suggests that Firmicute-assigned scaffolds are a contaminant.
SDAX01: 2 Gb in 2,688,687 scaffolds, N50 1,128 bp; 550,837 scaffolds (over 170 Mbp) have coverage below 0.01 in SRR1714990, of which 433,970 (over 136 Mb) are covered in SRR1719763 or SRR1712902; hiding no-hit scaffolds leaves 43,857 (1.6%) spanning 209 Mb (10.2%); full interactive view needs at least 8 GB RAM. · Visualisation of highly fragmented assemblies; Figure legends
ShownPublic eukaryotic assemblies at ENA lack explicit links to the read runs used to build them, which limits coverage-based contamination checks; this prompted new ENA guidance on referencing reads at submission, while separate work embedded BlobToolKit plots in ENA assembly pages.
Of 7,632 ENA eukaryotic assemblies associable with read sets, 585 (8%) linked to a single run, 875 (11%) to two to four runs and 6,172 (81%) to four or more; none explicitly cited their source runs in metadata; the 585 single-run assemblies cover 266 unique species. · ENA Integration; Discussion
Methods, data and reproducibility
BlobTools2 (python3) builds a per-contig JSON BlobDir from an assembly FASTA plus read alignments, BLAST/Diamond hits (taxrules bestsum, bestsumorder) and BUSCO output; a React/d3 browser Viewer with an Express API offers filterable blob, cumulative, snail, BUSCO and hit views, and a Snakemake pipeline screens INSDC assemblies.
- Public INSDC eukaryotic genome assemblies processed by INSDC-pipeline and served on the public BlobToolKit Viewer (snapshot of 13th November 2019) · multiple eukaryotic species
- ENA eukaryotic genome assemblies that could be associated with raw read sets (read-run linkage survey) · multiple eukaryotic species · n = 7,632 assemblies
- Drosophila albomicans assembly ACVV01 with read set SRR026696 (Acetobacter, Gluconobacter and Wolbachia signals) · Drosophila albomicans · GCA_000298335.1; SRR026696
- Drosophila elegans assembly AFFF02 used for BUSCO placement comparison · Drosophila elegans · AFFF02
- Conus consors assembly SDAX01, with the three largest of nine associated short-read libraries mapped · Conus consors (cone snail) · n = 2,688,687 scaffolds · GCA_004193615.1; PRJNA267645; SRR1714990, SRR1719763, SRR1712902
- Thicket tinamou assembly PTEZ01 with read set SRR6918124 · Crypturellus cinnamomeus · GCA_003342915.1; SRR6918124
- Northern bobwhite assembly AWGT02 screened for apicomplexan-assigned scaffolds · Colinus virginianus · GCA_000599465.2
- Sperm whale assemblies AWZP01 and chromosome-level RefSeq representative PGGR02, plus another recent assembly · Physeter catodon · GCA_000472045.1; GCA_002837175.2
- Methods & tools
- BlobTools2 v2.1 (python3 command-line tool, BlobDir JSON format), BlobToolKit Viewer v1.0 (React, Redux, reselect, d3), BlobToolKit Specification v1.0 (JSON-schema validator), Express Node.js API, NCBI BLAST+ blastn against NCBI nt, Diamond searches against UniProt, BUSCO 3.0.2 (e.g., diptera_odb9), pysam, minimap2, windowmasker, Snakemake INSDC-pipeline v1.0 with Conda environments, Selenium WebDriver
- Code
- https://github.com/blobtoolkithttps://github.com/blobtoolkit/blobtools2https://github.com/blobtoolkit/insdc-pipelinehttps://github.com/blobtoolkit/specificationhttps://github.com/blobtoolkit/viewer
- Data & accessions
- https://blobtoolkit.genomehubs.org/viewhttps://blobtoolkit.genomehubs.org/api-docs/https://blobtoolkit.genomehubs.org/downloadhttps://www.ebi.ac.uk/ena/browser/view/GCA_000298335https://www.ncbi.nlm.nih.gov/genome/24193
- Software & environment
- BlobTools2 v2.1, INSDC-pipeline v1.0, Specification v1.0, Viewer v1.0 archived on Zenodo; pipeline dependencies defined in Conda YAML files; git commit hash and dated database directories recorded in BlobDir metadata; BUSCO 3.0.2 used.
- Assay conditions & compute
- Interactive viewing of the 2,688,687-scaffold SDAX01 needs at least 8 GB RAM and a browser without RAM caps (e.g. Firefox); datasets over 1 million scaffolds use pre-generated static images. Pipeline hardware and runtime not reported.
- Reproducibility gaps
- Supplementary Files S1-S3 are said to be on Figshare but no link or DOI is printed; coverage is only added where read runs can be linked to an assembly; analysis of all public INSDC genomes was still in progress.
Limitations (6)
- Contig taxonomy from BLAST/Diamond similarity is provisional because reference databases contain mis-annotated records, as the tinamou case shows with apicomplexan sequence labelled as whale or quail. · stated by the authors
- Binned blob plots with one scaling scheme can downplay minor contig sets, making cobionts harder to spot, while scaled-circle plots of many contigs become hard to read because prominence depends on plotting order. · stated by the authors
- Assemblies with millions of contigs are impractical to explore interactively; the public Viewer serves static images and filters no-hit scaffolds above 1 million scaffolds, and interactive viewing of the 2,688,687-scaffold SDAX01 needed at least 8 GB RAM. · stated by the authors
- Coverage-based screening of public assemblies depends on inferring which reads built each assembly, since no ENA eukaryotic assembly explicitly referenced its source runs. · stated by the authors
- No quantitative benchmark (sensitivity, specificity or runtime) against other contamination screening or binning tools such as Anvi'o is presented; the evidence comes from selected case studies. · noted in this summary
- Case-study interpretations (Wolbachia nuclear insertions, apicomplexan contamination, questionable Conus scaffolds) rest on GC, coverage and similarity patterns without orthogonal validation such as PCR or long-read data. · noted in this summary
Who should read:Read this if you assemble or reuse eukaryotic genomes and need an interactive, reproducible way to spot contaminants, cobionts or parasite sequence using GC, coverage and taxonomic hits.
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
- Using QC-Blind for quality control and contamination screening of bacteria DNA sequencing data without reference genome
Xi Wang, Yan Gao, … Kang Ning·Frontiers in MicrobiologyPeer-reviewed·Jul 9, 2019
- Consensus assessment of the contamination level of publicly available cyanobacterial genomes
Luc Cornet, Loïc Meunier, … Denis Baurain·PLoS ONEPeer-reviewed·Jul 25, 2018
- Assessing species coverage and assembly quality of rapidly accumulating sequenced genomes
Romain Feron, Robert Michael Waterhouse·GigaSciencePeer-reviewed·Jan 1, 2022
- Automated ensemble assembly and validation of microbial genomes
Sergey Koren, Todd J Treangen, … Adam M Phillippy·bioRxiv·Feb 7, 2014
- GenomeQC: A quality assessment tool for genome assemblies and gene structure annotations
Nancy Manchanda, John L. Portwood, … Matthew B. Hufford·BMC GenomicsPeer-reviewed·Mar 2, 2020
- 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