Evaluating MAG Quality: What Completeness and Contamination Thresholds Should You Use?

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-21

Key Takeaways

• Metagenome-Assembled Genome (MAG) quality is primarily assessed by completeness and contamination, with the Minimum Information about a Metagenome-Assembled Genome (MIMAG) framework providing widely adopted thresholds: medium quality (≥50% complete, <10% contamination) and high quality (≥90% complete, <5% contamination).
• Completeness and contamination are inferred using single-copy marker genes, with tools like CheckM (lineage-specific markers) and BUSCO (universal orthologs) offering complementary estimates; discrepancies between tools necessitate manual inspection of MAGs.
• The appropriate quality thresholds are context-dependent, dictated by the research question; taxonomic profiling may suffice with medium-quality MAGs, while comparative genomics and metabolic reconstruction demand high-quality or near-complete MAGs (≥95% complete, <5% contamination).
• Beyond MIMAG thresholds, critical quality dimensions include chimerism (contigs from different organisms binned together), which can be assessed by tools like GUNC, and strain heterogeneity, which is not directly captured by standard completeness/contamination metrics.
• Reproducibility in MAG quality assessment requires meticulous documentation of tool versions, database versions, parameters, and decision-making processes, enabling downstream analyses and public database submissions.
• For applications involving pathogen identification, antimicrobial resistance surveillance, or agricultural/aquaculture management, stricter quality thresholds are paramount to mitigate the risks of false positives and ensure reliable biological interpretations.

Metagenome-assembled genomes (MAGs) are genomes reconstructed from shotgun metagenomic sequencing data by binning contigs that share coverage and composition signals. Researchers routinely ask what completeness and contamination thresholds make a MAG acceptable for publication, comparative genomics, or functional annotation. The Minimum Information about a Metagenome-Assembled Genome (MIMAG) framework provides the most widely used reference points: a medium-quality MAG should be at least 50% complete with less than 10% contamination, while a high-quality MAG should be at least 90% complete with less than 5% contamination. These thresholds are not arbitrary, but they are also not universal. The appropriate cutoffs depend on your research question, the tools you use to estimate quality, the complexity of your sample, and the downstream analyses you plan to run. This article explains how to interpret completeness and contamination estimates from CheckM and BUSCO, how to apply MIMAG standards in practice, and how to document your quality decisions for publication and public database submission.

At a Glance: MAG Quality Thresholds and Their Applications

The table below summarizes the MIMAG quality tiers and the practical contexts where each tier is appropriate. These values represent the consensus standards used across the metagenomics community and are the minimum expectations for most journals and public repositories.

The distinction between medium and high quality matters most when you are deciding whether a MAG can support claims about gene content or metabolic capacity. A MAG that is 60% complete may still be useful for confirming that a taxon is present in your sample, but it cannot reliably support the conclusion that an organism lacks a particular pathway. The absence of a gene in an incomplete MAG is not evidence of absence in the organism.

Understanding Completeness and Contamination Estimates

Completeness and contamination are not measured directly. They are inferred from the presence of single-copy marker genes that are expected to occur exactly once in a bacterial or archaeal genome. The logic is straightforward: if you recover most of the expected single-copy markers, your MAG is likely mostly complete. If you recover multiple copies of markers that should be single-copy, your MAG likely contains contigs from more than one organism.

How CheckM Estimates Quality

CheckM uses a set of lineage-specific marker genes that are identified by placing your MAG in a reference tree and selecting markers that are single-copy in the closest reference lineages. The tool then estimates completeness as the fraction of those markers that are present in your MAG and contamination as the fraction of markers that appear in multiple copies. This lineage-specific approach gives CheckM an advantage in accuracy over generic marker sets, but it also means that CheckM results depend on the reference genomes available for your taxonomic group. For poorly represented lineages, the marker set may be less informative and the estimates less reliable.

How BUSCO Estimates Quality

BUSCO (Benchmarking Universal Single-Copy Orthologs) uses a curated set of single-copy orthologs that are expected to be present in all genomes within a taxonomic group, such as bacteria or archaea. BUSCO reports completeness as the percentage of BUSCO groups that are complete and contamination as the percentage of BUSCO groups that are duplicated. The BUSCO approach is more standardized across lineages than CheckM, but it can be less sensitive for lineages that are distantly related to the reference genomes used to build the BUSCO sets. Many researchers report both CheckM and BUSCO estimates for their MAGs, and discrepancies between the two tools can indicate problems with the bin or with the reference databases.

Why Estimates Can Disagree

It is common to see CheckM and BUSCO produce different completeness and contamination values for the same MAG. These differences arise from the underlying marker sets and the algorithms used to detect them. CheckM markers are lineage-specific and may be more sensitive for well-represented taxa, while BUSCO markers are universal and may be more consistent across diverse lineages. When the two tools disagree substantially, you should investigate the MAG manually. Check the coverage profile of the contigs, look for contigs with anomalous GC content, and examine the taxonomic assignment of the bin. Disagreement between tools is a signal that the quality estimates are uncertain, not a reason to pick the more favorable number.

The MIMAG Standards in Context

The MIMAG standards were developed to provide a common language for describing MAG quality across studies. The thresholds for medium and high quality were chosen to balance practicality against rigor. A medium-quality MAG at 50% completeness is sufficient for many ecological questions, such as which taxa are present and how abundant they are. A high-quality MAG at 90% completeness is expected to support most genomic analyses, including gene content comparisons and metabolic pathway reconstruction.

What the Thresholds Do Not Tell You

The MIMAG thresholds do not address several important aspects of MAG quality. They do not measure chimerism, which occurs when contigs from two different organisms are binned together. They do not assess whether the MAG represents a single strain or a population of closely related strains. They do not evaluate the accuracy of the taxonomic assignment or the quality of the underlying assembly. A MAG can pass the MIMAG thresholds and still be problematic for certain analyses. For example, a MAG with 95% completeness and 2% contamination could contain a small number of contigs from a different species that carry antibiotic resistance genes, leading to false conclusions about the metabolic capabilities of the target organism.

When to Apply Stricter Thresholds

For some applications, the MIMAG thresholds are too permissive. If you are using MAGs to identify novel biosynthetic gene clusters, to predict antibiotic resistance, or to support clinical or industrial decisions, you should consider requiring higher completeness and lower contamination. Some researchers use a 95% completeness and 5% contamination cutoff for MAGs that will be used for detailed functional annotation. Others require that MAGs be manually curated to remove contaminating contigs before they are used for comparative genomics. The appropriate threshold depends on the cost of a false positive or false negative in your specific context.

Practical Workflow for MAG Quality Assessment

The workflow below describes the steps you should follow to assess MAG quality in a reproducible way. This workflow assumes you have already assembled your metagenome and generated bins using a tool such as MetaBAT, MaxBin, or CONCOCT.

Step 1: Run Quality Estimation Tools

Run CheckM and BUSCO on all of your bins. Use the same versions of the tools and the same databases for all samples in your study. Record the versions in your methods section so that other researchers can reproduce your results. The Galaxy Training Network provides accessible tutorials for running metagenomics workflows, including quality assessment, that can help you establish a reproducible pipeline.

Step 2: Compare Estimates Across Tools

Create a table that lists the completeness and contamination estimates from CheckM and BUSCO for each bin. Flag any bins where the two tools disagree by more than 10 percentage points for completeness or 5 percentage points for contamination. These bins require manual inspection before you decide whether to include them in your analysis.

Step 3: Assess Chimerism

Run a tool such as GUNC to detect chimerism in your MAGs. Chimeric MAGs can pass completeness and contamination thresholds while still containing sequences from multiple organisms. The nf-core documentation describes community pipelines that include chimerism checks as part of standard MAG quality assessment, and you can adapt these workflows to your own data.

Step 4: Manually Inspect Flagged Bins

For bins that fail quality thresholds or show tool disagreement, examine the coverage and GC content of the contigs. Plot coverage against GC content and look for contigs that deviate from the main cluster. Check the taxonomic assignment of individual contigs using a tool such as Kraken or CAT. Remove contigs that are clearly from a different organism and rerun the quality estimation.

Step 5: Assign Taxonomy

Classify your MAGs using a tool such as GTDB-Tk or a reference-based classifier. The taxonomic assignment should be consistent with the marker genes used for quality estimation. If your MAG is assigned to a lineage that is very different from the reference genomes used by CheckM, the quality estimates may be less reliable.

Step 6: Document Everything

Record the tool versions, database versions, and parameters used for every quality assessment step. This documentation is essential for reproducibility and is required by most journals. The Bioconductor project emphasizes reproducible genomic analysis workflows, and the same principles apply to metagenomics. Your methods section should allow another researcher to reproduce your exact quality assessment pipeline.

Options and Tradeoffs in Quality Estimation Tools

The choice of quality estimation tool can affect your results and your ability to compare your MAGs with those from other studies. Each tool has strengths and limitations that you should consider before committing to a pipeline.

CheckM versus BUSCO

CheckM is the most widely used tool for MAG quality estimation and is the default in many metagenomics pipelines. Its lineage-specific approach can be more accurate for well-represented taxa, but it requires a reference tree and can be slow for large datasets. BUSCO is faster and more standardized, but it may be less sensitive for lineages that are poorly represented in its reference sets. Many studies report both, and the EMBL-EBI training resources provide guidance on interpreting results from different quality assessment tools.

Single-Copy Marker Genes versus Coverage-Based Approaches

Some newer tools use coverage information in addition to marker genes to estimate MAG quality. These approaches can detect contamination that marker-based methods miss, particularly when the contaminating organism is closely related to the target organism. However, coverage-based methods require accurate mapping of reads to the MAG, which can be problematic for repetitive regions or for MAGs with uneven coverage. The tradeoff is between sensitivity and computational cost.

The Role of Reference Databases

All quality estimation tools depend on reference databases. CheckM uses a reference tree built from genomes in GenBank, while BUSCO uses ortholog sets built from representative genomes. The NCBI data resources provide the underlying sequence data for these databases, and the quality of your estimates depends on the quality and completeness of these references. For novel or poorly represented lineages, you should interpret quality estimates with caution and consider supplementing them with manual inspection.

Records and Measurements for MAG Quality

Keeping detailed records of your quality assessment is essential for publication, for public database submission, and for your own ability to revisit and reinterpret your results. The following records should be maintained for every MAG in your study.

Essential Records

For each MAG, record the bin identifier, the assembly and binning tools used, the quality estimation tools and versions, the completeness and contamination estimates from each tool, the taxonomic assignment, and the final quality tier assigned. Also record the number of contigs, the total length, the N50, and the GC content of the MAG. These basic statistics provide context for the quality estimates and can help you identify problems.

Sample-Level Records

Record the sequencing depth, the number of reads, and the assembly statistics for each sample. MAG quality is influenced by sequencing depth and community complexity, and these sample-level factors should be reported alongside the MAG-level quality estimates. The Galaxy Training Network provides templates for documenting metagenomics workflows that you can adapt to your study.

Database Submission Records

If you plan to submit your MAGs to a public database such as NCBI, you will need to provide quality estimates as part of the submission. The NCBI data resources describe the required metadata for genome submissions, and you should ensure that your quality records are formatted to meet these requirements. Incomplete or inconsistent quality records can delay or prevent database submission.

Common Failure Patterns in MAG Quality Assessment

Understanding the common ways that MAG quality assessment goes wrong can help you avoid these problems in your own work. The patterns below are frequently observed in metagenomics studies and are worth checking in your own data.

Overestimating Completeness

Completeness can be overestimated when the marker genes used for estimation are present in multiple copies in the target organism. Some lineages have naturally duplicated marker genes, and if the reference database does not account for these duplications, the completeness estimate will be inflated. This problem is more common in lineages that are distantly related to the reference genomes used to build the marker sets.

Underestimating Contamination

Contamination can be underestimated when the contaminating organism is closely related to the target organism. If two strains of the same species are binned together, the marker genes may appear as single copies because the two strains share the same markers. This problem is particularly acute in samples with high strain diversity, such as gut microbiomes or environmental samples with closely related populations.

Ignoring Chimerism

Chimeric MAGs can pass completeness and contamination thresholds while containing sequences from multiple organisms. Chimerism is most common in complex communities where coverage varies across organisms. A chimeric MAG can lead to false conclusions about gene content, metabolic capacity, or taxonomic assignment. Always check for chimerism, especially in MAGs that will be used for detailed functional analysis.

Tool Disagreement Without Investigation

When CheckM and BUSCO disagree, some researchers simply report the more favorable estimate or average the two values. This practice obscures the uncertainty in the quality estimates and can lead to incorrect conclusions. Tool disagreement should trigger manual inspection, not averaging.

Applying Thresholds Without Context

Applying the same completeness and contamination thresholds to all MAGs regardless of their intended use can lead to problems. A MAG that is adequate for taxonomic profiling may be inadequate for metabolic reconstruction. The thresholds should be chosen based on the downstream analyses and the cost of errors in your specific context.

Limitations of Completeness and Contamination Estimates

Completeness and contamination estimates are useful but imperfect measures of MAG quality. Understanding their limitations is essential for interpreting your results and for communicating uncertainty to other researchers.

Estimates Are Inferred, Not Measured

Completeness and contamination are inferred from marker gene presence and copy number. These inferences depend on the accuracy of the marker sets and the reference databases. For lineages that are poorly represented in reference databases, the estimates can be substantially wrong. The EMBL-EBI training resources provide background on how marker-based quality estimation works and where it can fail.

Estimates Do Not Capture All Quality Dimensions

Completeness and contamination do not capture all aspects of MAG quality. They do not measure base-level accuracy, assembly correctness, or the presence of misassembled contigs. A MAG can have high completeness and low contamination while still containing assembly errors that affect gene prediction and functional annotation. For critical applications, you should consider additional quality checks, such as mapping reads back to the MAG and checking for coverage uniformity.

Estimates Are Relative to Reference Databases

The quality estimates are relative to the reference databases used by the tools. As reference databases improve, the estimates for the same MAG can change. This means that quality estimates are not directly comparable across studies that use different database versions. Always report the database versions used in your analysis.

Strain Heterogeneity Is Not Captured

MAGs often represent populations of closely related strains instead of single genomes. The completeness and contamination estimates do not capture this strain heterogeneity. A MAG that appears to be high quality may actually contain sequences from multiple strains, which can affect analyses that assume a single genome, such as variant calling or gene content comparison.

Welfare and Safety Context for MAG Quality Assessment

While MAG quality assessment is a computational task, it has implications for biological safety and for the interpretation of research involving pathogens or clinically relevant organisms. The following considerations are relevant for researchers working with MAGs from clinical, agricultural, or environmental samples.

Pathogen Identification and Public Health

MAGs are increasingly used to identify pathogens in clinical and environmental samples. A low-quality MAG can lead to false identification of a pathogen or to incorrect conclusions about virulence or antibiotic resistance. The study of Campylobacter MAGs at a wildlife-livestock-human interface in Uganda demonstrates how MAGs can be used to identify novel species and to assess their clinical significance. In such studies, the quality thresholds should be set to minimize false positives, because a false pathogen identification can have serious public health consequences.

Antimicrobial Resistance Surveillance

MAGs are used to track antimicrobial resistance genes in environmental and clinical samples. The metagenomic characterization of infected diabetic foot ulcers in North Africa identified a high-quality Pseudomonas aeruginosa MAG carrying 220 virulence factors and 60 antibiotic resistance genes. The quality of this MAG was critical for the reliability of the resistance gene predictions. If the MAG had been contaminated with sequences from other organisms, the resistance gene profile would have been misleading.

Agricultural and Aquaculture Applications

MAGs are used to study microbial communities in agricultural and aquaculture systems. The study of 403 MAGs from kelp farming regions used medium- to high-quality MAGs to characterize the microbial community and to identify potential biomarkers for disease. In such applications, the quality thresholds should be chosen based on the specific management decisions that will be informed by the results. A MAG used to guide disease management decisions should meet the high-quality threshold.

Professional Escalation Criteria

If you are working with MAGs that will inform clinical, agricultural, or environmental management decisions, you should escalate to a specialist when the quality estimates are uncertain or when the MAGs fail to meet the thresholds required for your application. A bioinformatics specialist or a metagenomics methodologist can help you interpret the quality estimates and decide whether additional sequencing or manual curation is needed. Do not proceed with downstream analyses that depend on MAG quality when the quality is uncertain.

Building a MAG Quality Decision Framework for Your Specific Research Question

The MIMAG thresholds provide a useful starting point, but they do not tell you which threshold is right for your particular study. A decision framework that maps your research question to specific quality requirements will save you from two common mistakes: rejecting useful MAGs because they do not meet an unnecessarily strict cutoff, or including problematic MAGs because they pass a threshold that is too permissive for your intended analysis. This section provides a structured approach for setting quality cutoffs based on your downstream goals, a record system for tracking quality decisions, and troubleshooting methods for MAGs that fall into ambiguous quality territory.

Mapping Research Questions to Quality Thresholds

The first step in any MAG quality assessment is to define what you will do with the MAG after it passes quality control. Different downstream analyses have different tolerances for incompleteness and contamination, and your thresholds should reflect those tolerances instead of a one-size-fits-all standard.

Taxonomic Presence and Abundance Surveys

If your goal is to determine which organisms are present in a sample or to compare community composition across samples, medium-quality MAGs at 50% completeness and less than 10% contamination are often sufficient. These MAGs can confirm the presence of a taxon and support relative abundance estimates based on read mapping. The kelp farming study that reconstructed 403 MAGs used medium- to high-quality MAGs to characterize microbial communities and identify core taxa across sampling sites. For presence and absence questions, the cost of a false negative is higher than the cost of a false positive, so you should prioritize completeness over contamination control. A MAG that is 60% complete will still detect the presence of a dominant taxon, but it may miss low-abundance organisms that are only partially recovered.

Gene Content and Metabolic Reconstruction

When you move from asking which organisms are present to asking what those organisms can do, your quality requirements increase substantially. Gene content analysis and metabolic pathway reconstruction require high-quality MAGs at 90% completeness and less than 5% contamination. The reason is straightforward: the absence of a gene in an incomplete MAG is not evidence of absence in the organism. If your MAG is only 80% complete, you will miss approximately 20% of the genes, and any conclusion about the presence or absence of a metabolic pathway becomes unreliable. The diabetic foot ulcer study that reconstructed a Pseudomonas aeruginosa MAG reported a MAG at 99.68% completeness and 0.89% contamination before profiling virulence factors and antibiotic resistance genes. The authors did not rely on a minimum threshold, they achieved near-complete recovery to ensure that their resistance gene predictions were trustworthy.

Comparative Genomics and Phylogenomics

Comparative genomics, including pangenome analysis, synteny assessment, and phylogenetic tree construction, demands the highest quality MAGs. For these analyses, you should require at least 90% completeness and preferably 95% or higher, with contamination below 5%. The Campylobacter study at a wildlife-livestock-human interface reconstructed 44 MAGs representing seven species, including three novel species, and used these for comparative genomic analysis. Novel species descriptions and phylogenetic placement require near-complete genomes because missing genes can distort tree topology and lead to incorrect taxonomic conclusions. If you are describing a new species, you should aim for the near-complete tier and consider manual curation to remove any contaminating contigs.

Functional Annotation and Pathway Prediction

Functional annotation, including carbohydrate-active enzyme profiling, secondary metabolite biosynthetic gene cluster prediction, and virulence factor identification, requires high-quality MAGs but with a different emphasis than comparative genomics. For functional annotation, contamination is more dangerous than incompleteness. A small amount of contamination from another organism can introduce foreign genes that you will incorrectly attribute to your target organism. The integrative meta-omics study in Galaxy used MAGs to analyze carbohydrate-active enzymes and metabolic pathways in a cellulose-degrading consortium. The authors emphasized MAG quality because functional annotations are only as reliable as the genomes they are based on. For functional annotation, you should require less than 5% contamination and consider using a stricter cutoff of 2% if your MAGs will be used to predict antibiotic resistance or virulence factors.

Setting Your Quality Cutoffs: A Practical Decision Procedure

The following procedure will help you set quality cutoffs that are appropriate for your specific research question. This procedure assumes you have already run CheckM and BUSCO on all of your bins and have the results in a table.

Step 1: Define Your Primary Downstream Analysis

Write down the single most important analysis you will perform with your MAGs. This analysis determines your minimum quality tier. If you are doing multiple analyses, use the most stringent requirement. For example, if you are using MAGs for both taxonomic profiling and metabolic reconstruction, you should use the metabolic reconstruction threshold of 90% completeness and less than 5% contamination.

Step 2: Determine Your Error Tolerance

Consider the cost of a false positive and a false negative in your specific context. A false positive occurs when you include a MAG that is contaminated and attribute foreign genes to your target organism. A false negative occurs when you exclude a MAG that is genuinely useful because it does not meet your completeness threshold. For clinical or industrial applications, false positives are more dangerous because they can lead to incorrect conclusions about pathogenicity or metabolic capability. For ecological surveys, false negatives are more problematic because they can lead to underestimating diversity.

Step 3: Select Your Primary Quality Tool

Choose whether CheckM or BUSCO will be your primary quality estimation tool. This choice should be based on your taxonomic group of interest. For well-represented lineages, CheckM is often more accurate because it uses lineage-specific markers. For diverse communities spanning multiple phyla, BUSCO provides a more consistent standard. The EMBL-EBI training resources provide guidance on selecting and interpreting quality assessment tools. Whichever tool you choose, report both estimates in your methods and explain why you prioritized one over the other.

Step 4: Apply Your Thresholds and Flag Ambiguous MAGs

Apply your chosen thresholds to all of your bins. Flag any MAG that falls within 5 percentage points of your completeness or contamination cutoff. These borderline MAGs require additional scrutiny before you decide whether to include them. For example, if your completeness threshold is 90%, flag any MAG between 85% and 95% complete. These MAGs may be usable, but you need to examine them more closely to determine whether the quality estimates are reliable.

Step 5: Manually Review Flagged MAGs

For each flagged MAG, examine the coverage and GC content of the contigs. Plot coverage against GC content and look for contigs that deviate from the main cluster. Check the taxonomic assignment of individual contigs using a tool such as Kraken or CAT. If the flagged MAG has a small number of contaminating contigs, you may be able to remove them and rerun the quality estimation. If the flagged MAG is incomplete because of low sequencing depth, consider whether additional sequencing is feasible or whether the MAG can still support your primary analysis.

Step 6: Document Your Decisions

Record the rationale for including or excluding each flagged MAG. This documentation is essential for reproducibility and for defending your quality decisions during peer review. The nf-core documentation emphasizes the importance of transparent and reproducible workflows, and your quality decision log should be part of your project records.

A Record System for MAG Quality Decisions

A structured record system will help you track quality decisions across your project and ensure that your methods are reproducible. The following record format can be adapted to your specific needs.

MAG-Level Quality Record

For each MAG, maintain a record with the following fields: bin identifier, assembly tool and version, binning tool and version, CheckM version and database, BUSCO version and lineage dataset, CheckM completeness and contamination, BUSCO completeness and duplication, GUNC chimerism score if run, taxonomic assignment from GTDB-Tk or another classifier, number of contigs, total length, N50, GC content, and the final quality tier assigned. Also record whether the MAG passed or failed your quality thresholds and the reason for any manual curation.

Sample-Level Quality Record

For each sample, record the sequencing platform, sequencing depth, number of reads, assembly tool and version, and assembly statistics. MAG quality is influenced by these sample-level factors, and reporting them alongside MAG-level quality estimates provides important context. The Galaxy Training Network provides templates for documenting metagenomics workflows that you can adapt to your study.

Decision Log

Maintain a separate log that records the rationale for any quality-related decisions. This log should include the date, the MAG identifier, the decision made, the reason for the decision, and any supporting evidence such as coverage plots or taxonomic classifications. This log is particularly important for borderline MAGs that you include or exclude based on manual inspection.

Troubleshooting MAGs That Fall Below Quality Thresholds

When your MAGs do not meet your quality thresholds, you have several options beyond simply excluding them from your analysis. The following troubleshooting methods can help you improve MAG quality or determine whether a MAG is usable despite falling below your initial cutoff.

Improving Completeness Through Co-Assembly

If your MAGs are incomplete because of low sequencing depth, consider performing a co-assembly of multiple samples from the same environment. The integrative meta-omics study implemented a metagenomics workflow that combined single- and co-assembly followed by dereplication to improve MAG quality in complex samples. Co-assembly can increase the coverage of individual genomes by combining reads from multiple samples, which can improve completeness. However, co-assembly can also introduce chimerism if the same organism is represented by different strains across samples, so you should check for chimerism after co-assembly.

Removing Contamination Through Manual Curation

If your MAGs have contamination above your threshold, you can attempt to remove contaminating contigs manually. Start by examining the coverage and GC content of individual contigs. Contigs that deviate from the main cluster in a coverage-versus-GC plot are candidates for removal. You can also check the taxonomic assignment of individual contigs using a tool such as Kraken or CAT. Remove contigs that are clearly from a different organism and rerun the quality estimation. This process is labor-intensive but can rescue MAGs that are otherwise close to your quality thresholds.

Adjusting Binning Parameters

If your bins have consistently high contamination, consider adjusting your binning parameters or using a different binning tool. Different binning tools use different algorithms and may produce different results for the same assembly. You can also combine results from multiple binning tools using a tool such as DAS Tool or MetaWRAP. The Bioconductor project provides reproducible genomic-analysis workflows that can help you implement a multi-tool binning strategy.

Increasing Sequencing Depth

If your MAGs are incomplete because of low coverage, the most direct solution is to increase sequencing depth. This approach is expensive and may not be feasible for all projects, but it is the most reliable way to improve completeness. Before committing to additional sequencing, consider whether your current MAGs are sufficient for your primary analysis. A MAG that is 70% complete may still be useful for taxonomic profiling, even if it is not suitable for metabolic reconstruction.

Accepting Lower Quality for Exploratory Analyses

For exploratory analyses, such as initial community characterization or hypothesis generation, you may choose to include lower-quality MAGs with the understanding that they will be validated later. If you take this approach, clearly distinguish exploratory MAGs from those used for confirmatory analyses in your methods and results. The kelp farming study provides an example of how medium-quality MAGs can be used for community-level analyses while reserving high-quality MAGs for more detailed genomic characterization.

Common Failure Patterns in Threshold Selection

Understanding how threshold selection goes wrong can help you avoid these problems in your own work.

Using a Single Threshold for All Analyses

Applying the same completeness and contamination thresholds to all MAGs regardless of their intended use is a common mistake. A MAG that is adequate for taxonomic profiling may be inadequate for metabolic reconstruction. Your thresholds should be tailored to your specific downstream analyses, and you should document the rationale for your choices.

Choosing Thresholds Based on What Other Studies Report

It is tempting to copy thresholds from published studies, but these thresholds may not be appropriate for your data or your research question. The appropriate thresholds depend on your sample complexity, sequencing depth, and downstream analyses. Use published thresholds as a starting point, but adjust them based on your specific context.

Ignoring Tool Disagreement

When CheckM and BUSCO disagree, some researchers simply report the more favorable estimate or average the two values. This practice obscures the uncertainty in the quality estimates and can lead to incorrect conclusions. Tool disagreement should trigger manual inspection, not averaging.

Setting Thresholds After Seeing the Results

Setting your quality thresholds after you have already examined your quality estimates introduces bias. You may unconsciously choose thresholds that include the MAGs you want to keep and exclude those you want to discard. Set your thresholds before you examine your quality estimates, and document the rationale for your choices in your methods.

Professional Escalation Criteria for Quality Decisions

If you are uncertain about whether a MAG meets the quality requirements for your application, you should escalate to a specialist. A bioinformatics specialist or a metagenomics methodologist can help you interpret quality estimates, troubleshoot problematic MAGs, and decide whether additional sequencing or manual curation is needed. You should escalate when you encounter any of the following situations: your MAGs consistently fail to meet quality thresholds despite troubleshooting, CheckM and BUSCO disagree substantially for multiple MAGs, your MAGs will be used for clinical or industrial decisions, or you are describing novel species and need to ensure that your genomes meet the highest quality standards. The NCBI data resources provide access to reference genomes and taxonomic databases that can help you interpret quality estimates, and the EMBL-EBI training resources offer guidance on advanced metagenomics analysis. Do not proceed with downstream analyses that depend on MAG quality when the quality is uncertain.

Frequently Asked Questions

What is the difference between a medium-quality and a high-quality MAG?

A medium-quality MAG is at least 50% complete with less than 10% contamination. A high-quality MAG is at least 90% complete with less than 5% contamination. The distinction matters because a medium-quality MAG is sufficient for taxonomic profiling and presence/absence surveys, while a high-quality MAG is expected to support comparative genomics and functional annotation. The MIMAG standards provide the formal definitions, and most journals and public databases use these thresholds as minimum requirements.

Why do CheckM and BUSCO give different completeness and contamination estimates for the same MAG?

CheckM uses lineage-specific marker genes identified from a reference tree, while BUSCO uses universal single-copy orthologs. The two tools can disagree when the reference databases are incomplete for your taxonomic group or when the MAG contains sequences from multiple organisms. Disagreement between tools is a signal that the quality estimates are uncertain and that you should manually inspect the MAG before proceeding with downstream analyses.

Can a MAG with high completeness and low contamination still be problematic?

Yes. Completeness and contamination estimates do not capture all aspects of MAG quality. A MAG can pass the MIMAG thresholds while containing assembly errors, chimeric contigs, or sequences from closely related strains. For critical applications, you should check for chimerism, map reads back to the MAG to assess coverage uniformity, and manually inspect contigs that deviate from the expected coverage and GC content.

What completeness and contamination thresholds should I use for MAGs that will be used for metabolic reconstruction?

For metabolic reconstruction, you should use the high-quality threshold of at least 90% completeness and less than 5% contamination. Some researchers require even higher completeness, such as 95%, for MAGs that will be used to predict the presence or absence of metabolic pathways. The absence of a gene in an incomplete MAG is not evidence of absence in the organism, so higher completeness reduces the risk of false negative conclusions.

How should I report MAG quality in my publications?

Report the tool versions, database versions, and parameters used for quality estimation. Report the completeness and contamination estimates from each tool you used, and state the quality tier assigned to each MAG. The Galaxy Training Network and the nf-core documentation provide examples of reproducible metagenomics workflows that include quality reporting. Your methods section should allow another researcher to reproduce your exact quality assessment pipeline.

What should I do if my MAGs do not meet the MIMAG thresholds?

If your MAGs do not meet the MIMAG thresholds, you have several options. You can improve the assembly by using different assembly parameters or by performing co-assembly of multiple samples. You can improve the binning by using a different binning tool or by combining results from multiple tools. You can manually curate the bins to remove contaminating contigs. You can also increase sequencing depth to improve the completeness of the MAGs. The study of integrative meta-omics workflows in Galaxy describes an optimized metagenomics workflow that combines single- and co-assembly with dereplication to improve MAG quality in complex samples.

Are the MIMAG thresholds appropriate for all types of samples?

The MIMAG thresholds are general guidelines, and the appropriate thresholds depend on your research question and the complexity of your sample. For simple communities with low diversity, you may be able to achieve higher quality MAGs than the MIMAG thresholds. For complex communities with high diversity, such as soil or gut microbiomes, you may need to accept lower quality MAGs or to focus on the most abundant taxa. The thresholds should be chosen based on the downstream analyses and the cost of errors in your specific context.

How do I know if my MAG is chimeric?

Chimerism occurs when contigs from two different organisms are binned together. You can detect chimerism by running a tool such as GUNC, which compares the taxonomic assignment of individual contigs within a MAG. You can also detect chimerism by examining the coverage and GC content of the contigs. Contigs that deviate from the main cluster in a coverage-versus-GC plot may be from a different organism. Chimeric MAGs can pass completeness and contamination thresholds, so chimerism checks are essential for MAGs that will be used for detailed functional analysis.

Related Bioinformatics Guides

• Genomic Data Analysis Tools: A Comparative Guide for Researchers
• Evaluating Genome Assembly Quality: Metrics and Tools
• Binning in Metagenomics: From Contigs to Genomes
• Metabolomics Data Analysis in R: A Practical Workflow
• Microbiome Data Analysis in R: A Practical Guide for Compositional Data

Related Clinical & Scientific Guides

• A Practical Guide to Detecting Antimicrobial Resistance Genes in Shotgun Metagenomic Data
• Computational Immunology: Modeling the Immune System
• How to Set Hard Filters for Germline Variant Calling: A Practical Guide to GATK Best Practices

References and Further Reading

• NCBI Data Resources. National Center for Biotechnology Information.
• EMBL-EBI Training. European Bioinformatics Institute.
• Bioconductor. Bioconductor Project.
• Galaxy Training Network. Galaxy Project.
• nf-core Documentation. nf-core.
• The Carpentries Lessons. The Carpentries.
• EULAR recommendations for cardiovascular risk management in rheumatic and musculoskeletal diseases, including systemic lupus erythematosus and antiphospholipid syndrome.. Annals of the rheumatic diseases, 2022.
• [m-Cyano-p-[(18)F]fluorohippurate.](https://pubmed.ncbi.nlm.nih.gov/23193621). 2004.
• Diagnosis and treatment of Hymenoptera venom allergy: S2k Guideline of the German Society of Allergology and Clinical Immunology (DGAKI) in collaboration with the Arbeitsgemeinschaft für Berufs- und Umweltdermatologie e.V. (ABD), the Medical Association of German Allergologists (AeDA), the German Society of Dermatology (DDG), the German Society of Oto-Rhino-Laryngology, Head and Neck Surgery (DGHNOKC), the German Society of Pediatrics and Adolescent Medicine (DGKJ), the Society for Pediatric Allergy and Environmental Medicine (GPA), German Respiratory Society (DGP), and the Austrian Society for Allergy and Immunology (ÖGAI).. Allergologie select, 2023.
• Integrative meta-omics in Galaxy and beyond.. Environmental microbiome, 2023.
• AI in Psoriatic Disease: Scoping Review.. JMIR dermatology, 2024.
• Rasch-Built Overall Disability Scale for IgM-Associated Polyneuropathy With and Without Anti-MAG Antibodies: IgM-RODS.. Journal of the peripheral nervous system : JPNS, 2025.
• Uterine distension media for outpatient hysteroscopy.. The Cochrane database of systematic reviews, 2021.
• Mental health in children and adolescents with overweight or obesity.. BMC public health, 2023.
• Development and Characterization of Intravenous Nanoemulsions Loaded with &lt,i&gt,Magnolia officinalis&lt,/i&gt, Neolignans.. 2026.
• Genome-resolved metagenomics of Campylobacter at a wildlife-livestock-human interface in Uganda reveals novel species. 2026.
• Decoding a Microbial Community for Healthy Kelp: 403 MAGs from the World's Largest Kelp Farming Region.. 2026.
• Metagenomic characterization of infected diabetic foot ulcers in North Africa: microbial diversity, virulome, and resistome profiling.. 2026.
• Environmental quality standards for barium in surface water : Proposal for an update according to the methodology of the Water Framework Directive. 2020.
• Water quality standards based on human fish consumption : Background document for revision of the WFDmethodology. 2016.
• Evaluation of Weld Defects in Petroleum Pipelines Using X-Ray for MAG, TIG, and SMAW Techniques. International Science and Technology Journal, 2024.
• Secondary Standards in the UKIRT Faint Standard Fields. Astrophysical Journal Supplement Series, 2025.
• MODELLING INFLUENCE OF HEAT INPUT ON FILLET WELD GEOMETRY DURING MAG WELDING IN DIFFERENT POSITIONS. Zbornik radova, 2026.
• Water quality standards related to human exposure in the Water Framework Directive : Considerations on fish consumption and swimming. 2013.
• Microstructural Evolution and ISO-Based Weld Quality in MAG and Laser Welding of HC420LA Steel Under Different Heat Inputs. Crystals, 2026.
• Standard-compliant detection of fillet weld surface imperfections for MAG-welding using a 3D-line scanner. Proceedings of SPIE the International Society for Optical Engineering, 2021.
• Hybrid welding (Laser-electric arc mag) of high yield point steel s960ql. Materials, 2021.
• Preparation, radiochemical purity control and stability of 99mTc-mertiatide (Mag-3). Annals of Nuclear Medicine, 2005.
• Development of Nd:YAG laser and laser/MAG hybrid welding for land pipeline applications. Welding and Cutting, 2004.

This article is educational and does not replace validated analysis plans, institutional policy, clinical interpretation, or specialist review.