Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
Short answer
QUAST (QUality ASsessment Tool) is an open-source bioinformatics tool that evaluates and compares genome assemblies. It calculates reference-free contiguity statistics (such as contig counts, total length, and N50) and reference-based correctness metrics (such as NA50, structural misassemblies, and genome fraction). It produces HTML summary reports, metric plots, and browser-based contig alignments via its bundled Icarus visualizer.
Definition and scope
Introduced by Gurevich et al. in 2013, QUAST standardizes quality control across genome assemblies. De novo assemblers often generate structural misassemblies, chimeric joins, or base errors due to genomic repeats, uneven coverage, and algorithmic heuristics. QUAST provides objective, automated benchmarking across multiple assemblies simultaneously.
QUAST operates across two primary analytical modes:
- Reference-free evaluation: When no reference is provided, QUAST measures contig length distributions, total assembled bases, GC content, and length-based contiguity including N50 and L50.
- Reference-based evaluation: When an annotated or unannotated reference is provided, QUAST aligns contigs using NUCmer or minimap2 to calculate genome fraction, duplication ratios, mismatch rates, indels, and structural misassemblies.
Before assembly and QUAST evaluation, sequencing workflows perform adapter trimming using fastp read trimming and filtering and assess raw sequence quality with FastQC raw read quality control.
How it works or how it is used in practice
QUAST runs via Python on the command line. Users supply draft FASTA contigs alongside an optional reference genome:
quast.py contigs.fasta -r reference.fasta -o quast_results/
Supplying multiple assembly files compiles a side-by-side comparative matrix in HTML, plain text, and tab-delimited formats, facilitating direct benchmarking across different assemblers or parameter choices.
QUAST evaluates assemblies through several core computational steps:
- Contig thresholding: By default, QUAST reports summary statistics on contigs ≥ 500 bp (configured via
--min-contig). It also tabulates counts and lengths at thresholds of ≥ 1,000 bp, ≥ 5,000 bp, ≥ 10,000 bp, ≥ 25,000 bp, and ≥ 50,000 bp. - Contiguity calculation: QUAST sorts contigs in descending length to calculate N50 (the contig length where cumulative length reaches 50% of the total assembly) and L50 (the minimum number of contigs covering 50% of the assembly). When a reference is provided, NG50 and LG50 use the reference genome length as the denominator.
- Structural misassembly detection: QUAST maps contigs to the reference to detect misassemblies. An extensive misassembly is recorded when flanking sequences align > 1 kbp apart on the reference (a relocation), align to opposite strands of the same chromosome (an inversion), or align to different chromosomes (a translocation). QUAST-LG raises this threshold to 7 kbp via
--extensive-mis-sizeto accommodate eukaryotic structural variations. - Alignment-adjusted contiguity (NA50 and NGA50): Chimeric joins can artificially inflate N50. To assess true structural contiguity, QUAST breaks contigs at misassembly breakpoints, discards unaligned segments, and calculates NA50 and NGA50 across the resulting aligned blocks.
- Area under the curve (auN and auNA): QUAST implements the auN and auNA metrics proposed by Heng Li (2020), calculating the area under the Nx and NAx curves to provide contiguity scores that are less sensitive to single large contigs.
- Interactive visualization: Outputs include cumulative length curves, GC distribution plots, and Icarus Contig Alignment Viewers highlighting misassemblies and unaligned regions along reference chromosomes.
For an in-depth breakdown of contig distribution metrics and their limitations, see the N50 assembly quality guide.
Types or related concepts
The QUAST ecosystem includes specialized modules for varying genome sizes and data complexities. The table below outlines the primary tools in the official toolkit:
| Module | Target Assembly Scope | Default Aligner | Extensive Misassembly Threshold | Core Capabilities |
|---|---|---|---|---|
| Standard QUAST (Gurevich et al., 2013) | Genomes < 100 Mbp (bacteria, fungi, viruses) | NUCmer / BLAST | 1 kbp | Reference-free and reference-based statistics, HTML/PDF reports, Icarus contig viewer |
| QUAST-LG (Mikheenko et al., 2018) | Large eukaryotic genomes (> 100 Mbp) | minimap2 | 7 kbp | Scalable memory usage, BUSCO completeness integration, reference-free k-mer completeness |
| MetaQUAST (Mikheenko et al., 2016) | Metagenomic communities and microbiomes | NUCmer / BLAST | 1 kbp | Multi-reference evaluation, automated NCBI reference downloads, interspecies translocations, Krona charts |
| Icarus (Mikheenko et al., 2016) | Interactive visualizer for draft assemblies | Integrated | N/A (visualizer) | Dynamic browser navigation of contig alignments, physical coverage, and structural breakpoints |
Researchers assembling multi-species communities can consult the metagenomics workflow guide for metagenome profiling and binning practices.
Where it matters for labs, clinics or supply decisions
Assembly quality assessment informs crucial computational and experimental decisions across sequencing laboratories:
- Assembler benchmarking: Core facilities evaluate assemblers (such as SPAdes, Flye, Canu, or MEGAHIT) across k-mer sizes to identify configurations that optimize contiguity while minimizing structural errors.
- Assessing sequencing platforms: Long-read platforms like Oxford Nanopore sequencing generate contigs with high N50 values but may have higher indel rates. QUAST quantifies base mismatch and indel frequencies per 100 kb, establishing whether short-read polishing is required.
- Preventing chimeric errors: Scaffolds with high N50 values can harbor chimeric joins. Designing PCR primers from misassembled contigs risks experimental failure and wasted reagents.
- Distinguishing biology from assembler error: Structural differences against a non-isogenic reference often reflect true biological variation rather than algorithmic defects. QUAST-LG enables structural variation masking via BEDPE files (
--sv-bedpe) to separate true variation from misassemblies. - Standardized data deposition: Public repositories like NCBI GenBank enforce assembly standards. QUAST verifies contig counts, scaffold gaps, and uncalled base (N) frequencies prior to submission.
Common misunderstandings
Key misunderstandings to avoid when interpreting QUAST output include:
- Assuming high N50 proves assembly accuracy: N50 measures length contiguity, not accuracy. Chimeric joins produce high N50 alongside elevated misassemblies. Comparing N50 with NA50 reveals whether contiguity reflects true structural correctness.
- Treating Duplication Ratio > 1.0 as high completeness: While genome fraction measures reference coverage, a duplication ratio well above 1.0 indicates that heterozygous alleles or repetitive elements failed to collapse, artificially inflating assembly size.
- Equating local misassemblies with extensive misassemblies: Extensive misassemblies represent major chromosomal rearrangements exceeding 1 kbp (7 kbp in QUAST-LG). Local misassemblies represent discrepancies between 200 bp and the extensive threshold, typically caused by small indels, repeat length variations, or alignment boundaries.
- Assuming unaligned contigs represent errors: In microbial isolates, unaligned contigs frequently correspond to plasmids, horizontal gene transfers, bacteriophages, or novel genomic islands not present in the reference strain.
- Running standard QUAST on mammalian genomes: Default QUAST consumes excessive RAM and runtime on multi-gigabase genomes when using NUCmer. Eukaryotic genomes should be evaluated using QUAST-LG via
--large, which switches to minimap2 and applies appropriate structural variation thresholds.
FAQ
What is the difference between N50 and NA50 in QUAST?
N50 is calculated purely from contig lengths, representing the contig size covering 50% of the assembly. NA50 is an alignment-adjusted metric: QUAST aligns contigs to a reference, splits them at every extensive misassembly breakpoint, removes unaligned bases, and recalculates N50 across the resulting aligned blocks to reflect true structural contiguity.
Can QUAST evaluate assemblies without a reference genome?
Yes. Without the -r flag, QUAST runs in de novo mode. It computes contig counts, largest contig length, total assembly size, GC content, N50, and L50. Alignment-dependent metrics like misassemblies, NA50, and genome fraction are omitted.
What aligner does QUAST use for reference mapping?
Standard QUAST and MetaQUAST use NUCmer (from MUMmer) for nucleotide alignments. QUAST-LG (activated with --large) uses minimap2, providing scalable alignment speed and manageable memory requirements for large eukaryotic genomes.
How does MetaQUAST handle unknown microbial species?
When community species are uncharacterized, MetaQUAST automatically searches NCBI via BLAST to download and align candidate reference genomes, providing per-species contiguity and alignment breakdowns.
What threshold defines extensive versus local misassemblies?
In standard QUAST, extensive misassemblies involve structural rearrangements or flanking gaps exceeding 1 kbp (7 kbp in QUAST-LG). Local misassemblies involve flanking gaps or overlaps between 200 bp and the extensive threshold on the same reference chromosome and strand.
Sources
- Gurevich, A., Saveliev, V., Vyahhi, N., & Tesler, G. (2013). QUAST: quality assessment tool for genome assemblies. Bioinformatics, 29(8), 1072–1075. DOI: 10.1093/bioinformatics/btt086 (read 2026-11-12). https://academic.oup.com/bioinformatics/article/29/8/1072/245229
- Mikheenko, A., Prjibelski, A., Saveliev, V., Antipov, D., & Gurevich, A. (2018). Versatile genome assembly evaluation with QUAST-LG. Bioinformatics, 34(13), i142–i150. DOI: 10.1093/bioinformatics/bty266 (read 2026-11-12). https://academic.oup.com/bioinformatics/article/34/13/i142/5045794
- Mikheenko, A., Saveliev, V., & Gurevich, A. (2016). MetaQUAST: evaluation of metagenome assemblies. Bioinformatics, 32(7), 1088–1090. DOI: 10.1093/bioinformatics/btv697 (read 2026-11-12). https://academic.oup.com/bioinformatics/article/32/7/1088/1744118
- Algorithmic Biology Lab (ABLab). (2024). QUAST 5.3.0 Manual and Documentation. Center for Algorithmic Biotechnology (read 2026-11-12). https://quast.sourceforge.net/docs/manual.html
- Li, H. (2020). A new metric on assembly contiguity: auN. Heng Li’s Blog (read 2026-11-12). https://lh3.github.io/2020/04/08/a-new-metric-on-assembly-contiguity







