- Genome completeness assessment is an essential part of genome assembly evaluation in bioinformatics. Genome assembly reconstructs genomic sequences from sequencing reads, but the resulting assembly may not contain every expected region of the original genome. Some genes or genomic regions may be missing, fragmented, duplicated, or incorrectly assembled. Completeness assessment helps researchers determine how much of the expected biological content has been recovered and whether an assembly is suitable for downstream analyses. One of the most widely used tools for this purpose is BUSCO, which evaluates genome completeness by searching for evolutionarily conserved genes expected to be present in a particular taxonomic group.
- BUSCO stands for Benchmarking Universal Single-Copy Orthologs. It is based on the principle that many genes have been conserved throughout evolution because they perform important biological functions. These genes are generally expected to occur in most members of a particular evolutionary lineage, often as single-copy genes. If an assembly contains most of the expected conserved genes, it provides evidence that a substantial proportion of the organism’s gene content has been recovered. BUSCO therefore provides a standardized approach for estimating gene-space completeness across genome assemblies, transcriptomes, and predicted protein datasets.
- The BUSCO approach uses sets of orthologous genes called BUSCO groups. These groups are organized into lineage-specific datasets representing different taxonomic levels, such as bacteria, archaea, fungi, plants, and animals. A lineage dataset contains genes selected because they are expected to be present in most organisms belonging to that lineage. The appropriate dataset depends on the organism being analysed. For example, a mammalian genome should be evaluated using a suitable mammalian lineage dataset, whereas a bacterial genome should use a relevant bacterial dataset. Selecting an inappropriate lineage can produce misleading completeness estimates because the expected gene content may not accurately represent the organism.
- BUSCO identifies expected genes by comparing sequences from the input data with the conserved ortholog groups in the selected lineage dataset. Depending on the input type, BUSCO may use genome sequence searches, gene prediction, or protein sequence comparisons. Detected BUSCO genes are classified into several categories that describe their completeness and copy number. These categories provide a summary of how much expected conserved gene content has been recovered and whether the detected genes occur in the expected single-copy or duplicated form.
- Complete single-copy BUSCOs represent conserved genes that have been detected as complete and occur once in the assembly or analysed dataset. These genes are consistent with the expected single-copy condition for the selected lineage. A high percentage of complete single-copy BUSCOs generally indicates good recovery of conserved gene content.
- Complete duplicated BUSCOs are conserved genes detected as complete but represented by more than one copy. Duplicated BUSCOs may indicate uncollapsed haplotypes, redundant sequence assembly, genome duplication, polyploidy, genuine gene duplication, or other biological features. They should not automatically be interpreted as assembly errors. In diploid and polyploid organisms, duplicated BUSCOs may reflect the way the assembler represents homologous chromosome copies. Their interpretation requires consideration of genome biology and assembly strategy.
- Fragmented BUSCOs are genes for which only part of the expected conserved sequence has been detected. Fragmentation may result from incomplete assembly, gaps, sequencing errors, incorrect gene prediction, highly divergent sequences, or interruptions in the assembled genomic region. A high proportion of fragmented BUSCOs can indicate that some expected genes have not been reconstructed completely. However, it may also reflect limitations in the input data or gene prediction methods.
- Missing BUSCOs are expected conserved genes that have not been detected in the analysed assembly or dataset. Their absence may indicate missing genomic regions, low sequencing coverage, assembly fragmentation, sequence divergence, or limitations in the selected lineage dataset. Some genes may also be genuinely absent from particular species because of evolutionary gene loss. Therefore, missing BUSCOs do not always indicate assembly failure.
- BUSCO completeness results are commonly summarized using the following categories:
C: Complete BUSCOs, including complete single-copy and complete duplicated genes.
S: Complete single-copy BUSCOs.
D: Complete duplicated BUSCOs.
F: Fragmented BUSCOs.
M: Missing BUSCOs.
n: Total number of BUSCO groups assessed in the selected lineage dataset.
The results are commonly reported in a compact format such as:
C:95.2% [S:91.0%, D:4.2%], F:2.1%, M:2.7%, n:954
- In this example, 95.2% of the expected BUSCO groups were detected as complete. Of these, 91.0% were complete single-copy genes and 4.2% were complete duplicated genes. Another 2.1% were fragmented, while 2.7% were missing. The analysis evaluated 954 BUSCO groups. This result suggests that the assembly contains a high proportion of the conserved gene content expected for the selected lineage, although it does not prove that the entire genome has been reconstructed accurately.
- The overall complete BUSCO percentage is calculated by adding complete single-copy and complete duplicated BUSCO percentages. Fragmented and missing BUSCO percentages account for the remaining categories. These values describe the recovery of expected conserved gene groups rather than the percentage of all genomic nucleotides recovered. A genome assembly can have a high BUSCO score while still missing large repetitive, intergenic, regulatory, or other noncoding regions. Conversely, an assembly may have a lower BUSCO score because of genuine biological differences or limitations in gene detection even when much of its genomic sequence has been recovered.
- BUSCO supports several types of input data. Genome mode evaluates assembled genomic DNA and may use gene prediction to identify coding regions. Transcriptome mode evaluates assembled transcripts, which may represent expressed genes rather than the complete genomic gene set. Protein mode evaluates predicted protein sequences and can be useful when a gene prediction pipeline has already produced a protein dataset. The choice of mode should match the biological data and the objective of the assessment.
- Genome mode is commonly used to evaluate genome assemblies because it searches for conserved genes directly in the assembled genomic sequence. This approach can provide information about the recovery of expected gene content while accounting for the challenges of identifying genes in genomic DNA. Transcriptome mode is useful for evaluating transcriptome assemblies, but its results depend on which genes were expressed in the sampled tissues, developmental stages, and environmental conditions. A transcriptome cannot be expected to contain every gene in the genome. Protein mode evaluates predicted protein sequences and is therefore influenced by the quality and completeness of the upstream gene prediction process.
- BUSCO is available through command-line software and can be integrated into genome assembly and annotation workflows. The following example illustrates a genome completeness assessment using BUSCO:
busco \
-i assembly.fasta \
-m genome \
-l mammalia_odb \
-o busco_genome_results \
-c 8
- In this command, -i specifies the input genome assembly, -m genome selects genome mode, -l identifies the lineage dataset, -o defines the output directory, and -c specifies the number of CPU cores. The lineage shown is an illustrative mammalian dataset choice. The appropriate lineage should be selected for the organism, and the exact dataset name and availability should be checked using the installed BUSCO version.
- BUSCO can also evaluate transcriptome assemblies. The following example illustrates transcriptome mode:
busco \
-i transcripts.fasta \
-m transcriptome \
-l eukaryota_odb \
-o busco_transcriptome_results \
-c 8
For predicted protein sequences, protein mode can be used:
busco \
-i proteins.fasta \
-m proteins \
-l eukaryota_odb \
-o busco_protein_results \
-c 8
- The lineage datasets in these examples are illustrative and should be replaced with the most appropriate available lineage for the organism. BUSCO versions may differ in their command-line options, dataset identifiers, and dependencies. Researchers should consult the documentation for their installed version and record the exact software and dataset versions used.
- Selecting the correct BUSCO lineage is an important part of completeness assessment. BUSCO datasets are available at different taxonomic levels, from broad groups to more specific evolutionary lineages. A broad lineage may be useful when the organism’s precise taxonomic placement is uncertain, but a more specific lineage can provide a better representation of the expected conserved genes when it is appropriate. Researchers should avoid selecting a lineage solely because it produces a higher completeness score. The dataset should reflect the organism’s evolutionary position and the biological purpose of the assessment.
- BUSCO results can be used to compare different genome assemblies of the same organism. For example, researchers may compare assemblies generated using short reads, long reads, or hybrid sequencing approaches. If one assembly has a higher complete BUSCO percentage and fewer fragmented or missing BUSCOs, it may have recovered more expected conserved genes. However, comparisons should also consider duplicated BUSCOs, assembly size, contamination, sequence accuracy, and structural correctness. A high complete BUSCO score alone is not sufficient to establish that one assembly is superior.
- BUSCO can also help evaluate genome polishing and annotation workflows. Researchers may run BUSCO before and after polishing to determine whether conserved gene recovery has changed. A substantial increase in complete BUSCOs may suggest improved reconstruction or gene detection, whereas an increase in fragmented or missing BUSCOs may indicate that the process introduced errors or affected gene regions. However, polishing should not be judged using BUSCO alone because many nucleotide-level corrections do not change whether a conserved gene is classified as complete. Read-based accuracy assessments and k-mer comparisons can provide complementary evidence.
- Completeness assessment is particularly important in metagenomics. Metagenome assembly reconstructs genomic sequences from mixed microbial communities, and the resulting metagenome-assembled genomes (MAGs) may represent only portions of individual microbial genomes. BUSCO can be used to evaluate the recovery of conserved genes in suitable taxonomic contexts, but microbial MAG assessment commonly uses tools such as CheckM and CheckM2. These tools estimate genome completeness and contamination using lineage-specific marker genes and other computational approaches. They are particularly useful for assessing whether a reconstructed microbial genome contains most expected genomic components and whether it may include sequences from other organisms.
- CheckM2 uses machine-learning-based approaches to estimate completeness and contamination in microbial genomes. It is useful for bacterial and archaeal genome quality assessment, including MAGs. CheckM and CheckM2 results should be interpreted in relation to the organism’s taxonomy, marker-gene representation, and genome characteristics. BUSCO and CheckM2 are complementary rather than interchangeable tools because they use different marker systems and analytical approaches. The choice depends on the type of genome and the biological question.
- Several additional approaches can complement BUSCO in genome completeness assessment. K-mer-based methods compare short sequence words found in raw sequencing reads with those present in the assembly. If many reliable read-derived k-mers are absent from the assembly, this may indicate missing genomic sequence or assembly errors. Merqury is a widely used tool for k-mer-based assembly evaluation and can provide information about completeness and consensus accuracy. These methods are particularly useful when high-quality sequencing reads are available.
- Reference-based assessment provides another source of evidence. If a closely related, high-quality reference genome is available, the assembled sequences can be aligned against it to identify missing regions, structural differences, and possible assembly gaps. Tools such as QUAST, MUMmer, and minimap2 can support these comparisons. However, reference-based assessment may be less reliable when the sample is genetically distant from the reference or when the reference itself contains errors or missing regions. Genuine biological differences should not automatically be interpreted as assembly defects.
- Genome completeness should also be considered alongside genome size and assembly continuity. The total assembly length can be compared with an independently estimated genome size, while metrics such as N50, L50, NG50, and contig count describe sequence continuity. An assembly may have a high BUSCO score but remain fragmented, particularly if conserved genes have been recovered in short contigs. Similarly, an assembly may have a high N50 but a low BUSCO score if large portions of expected gene content are missing. These metrics measure different properties and should be reported together.
- BUSCO has several important limitations. First, it evaluates conserved genes rather than every genomic region. It does not directly measure the completeness of intergenic regions, repetitive DNA, centromeres, telomeres, regulatory elements, or other noncoding sequences. Second, its results depend on the selected lineage dataset. Third, highly divergent genes may not be detected even if they are present. Fourth, genuine evolutionary gene loss may be classified as missing. Finally, duplicated BUSCOs can have biological or technical explanations, and their presence does not automatically indicate an assembly problem.
- Another limitation is that BUSCO results can be influenced by gene prediction. In genome mode, the software must identify gene regions from genomic DNA. If the assembly contains sequencing errors, unusual gene structures, incomplete exons, or highly divergent sequences, gene detection may be affected. Transcriptome and protein modes have their own limitations because they depend on expression patterns and upstream sequence prediction. Researchers should therefore document the input type and mode when reporting BUSCO results.
- Completeness assessment is especially valuable for high-quality reference genome development. In animal and plant genomics, recovering conserved genes is important for genome annotation, comparative genomics, gene-family analysis, and evolutionary studies. In microbial genomics, completeness is important for identifying metabolic capabilities, studying pathogenicity, and reconstructing microbial evolution. In metagenomics, it helps determine whether MAGs contain enough genomic information for downstream taxonomic and functional analyses. In comparative genomics, consistent completeness assessments help researchers select suitable genome resources and avoid comparisons based on highly incomplete assemblies.
- For reproducible analysis, researchers should record the BUSCO software version, lineage dataset, input file, analysis mode, computational parameters, and output summary. The same lineage dataset and software version should ideally be used when comparing assemblies. The final report should include complete single-copy, complete duplicated, fragmented, and missing BUSCO percentages, along with the total number of BUSCO groups assessed. Other relevant measures, such as assembly length, N50, contamination estimates, and k-mer-based accuracy, should also be provided.
- BUSCO is one of the most widely used tools for genome completeness assessment because it evaluates the recovery of evolutionarily conserved genes using standardized lineage-specific datasets. Its categories provide useful information about complete, duplicated, fragmented, and missing conserved genes. However, BUSCO measures gene-space completeness rather than the recovery of every nucleotide in the genome. A reliable genome quality assessment should therefore combine BUSCO with assembly continuity statistics, k-mer analysis, reference comparisons, contamination screening, and sequence accuracy evaluation. Using these complementary approaches provides a more comprehensive understanding of genome assembly quality and helps researchers select reliable genomic resources for downstream bioinformatics research.