# cgMLST Schemes in Bacterial Genomics ## Quick Answer - Select a cgMLST scheme that matches your organism, laboratory workflow, and surveillance purpose, then verify scheme version and locus composition before routine use. - Validate allele calling outputs against reference databases and authenticated control strains to confirm cluster assignments are reproducible across runs. - Interpret cluster thresholds with caution because schema choice and bioinformatics pipelines can change results more than biological variation for some species. ## What CgMLST Measures and Why It Matters for Bacterial Typing Core genome multilocus sequence typing, commonly abbreviated as cgMLST, extends the classical seven-gene multilocus sequence typing approach by comparing hundreds or thousands of core genes present across isolates of a bacterial species. Classical MLST examines a small set of housekeeping genes and provides a sequence type that is useful for broad population structure but often lacks the resolution needed for outbreak investigations. CgMLST instead indexes variation across the shared core genome, which gives laboratories a higher-resolution view of genetic relationships among isolates. The practical value of cgMLST lies in its ability to support source tracking, variant detection, and risk assessment during foodborne disease investigations. A systematic field investigation of Salmonella along a poultry production chain compared 12 typing methods and found that cgMLST, treated as the gold standard in that study, identified 18 distinct types among 61 isolates with a discrimination index of 0.8541. The same study showed that CRISPR-based multi-virulence locus sequence typing achieved a higher discrimination index of 0.9628 and detected 30 types, which illustrates that method choice depends on the specific surveillance question and available laboratory expertise. See the full comparison in the [Microbiology Spectrum study on subtyping methods for foodborne Salmonella surveillance](https://pubmed.ncbi.nlm.nih.gov/36194132). For laboratories moving from conventional typing to whole genome sequencing, cgMLST offers a bridge because it produces results that can be compared with historical MLST data. The Czech National Reference Laboratory for Streptococcal Infections implemented whole genome sequencing for invasive pneumococcal disease surveillance and compared isolates at three resolution levels: seven MLST genes, 53 ribosomal genes, and 1420 genes that included MLST genes, ribosomal genes, and cgMLST loci. This layered approach allowed the laboratory to maintain continuity with earlier MLST-based surveillance while gaining the higher resolution of a full core genome comparison. See the [Czech Republic implementation report on pneumococcal WGS surveillance](https://pubmed.ncbi.nlm.nih.gov/33086855). The choice of a cgMLST scheme is therefore not a trivial technical detail. It determines which loci are compared, how allele differences are counted, and ultimately how clusters are defined for public health action. ## Core Principles of CgMLST Scheme Design ### Definition of Core Genome Loci A cgMLST scheme begins with a definition of the core genome for a given bacterial species. The core genome consists of genes that are present in essentially all isolates of the species, as opposed to accessory genes that may be present in only some isolates. Scheme developers typically analyze a large collection of diverse isolates to determine which genes are conserved enough to be included. The number of loci in a cgMLST scheme varies substantially by species. The pneumococcal scheme used in the Czech Republic included 1420 loci when all currently defined genes in the PubMLST database were considered. A refined Neisseria gonorrhoeae scheme used 1430 core genes. For Campylobacter jejuni, different cgMLST schemas can differ by a factor of two in the number of loci, which has major consequences for cluster assignment. See the [inter-laboratory study on bioinformatic approaches for foodborne cluster detection](https://doi.org/10.3389/fmicb.2025.1629731). ### Allele Calling and Distance Calculation Once sequencing data are obtained, each isolate is assigned an allele number for every locus in the scheme. The allele numbers are determined by comparing the assembled or raw sequence at each locus against a reference database of known alleles. Isolates are then compared by calculating the number of allele differences between them, often expressed as an allele distance. The allele distance forms the basis for cluster definition. Isolates with allele distances below a threshold are considered part of the same cluster, which may indicate a common source in outbreak investigations. The threshold is typically species-specific and should be validated for each scheme. ### Scheme Versioning and Database Maintenance CgMLST schemes are living resources. As more isolates are sequenced, new alleles are discovered and added to reference databases. Scheme developers may also revise the set of core loci as more diverse isolates reveal that some genes are not truly core. Laboratories must track which scheme version was used for each analysis. Comparing isolates typed with different scheme versions can produce misleading results because the locus composition and allele definitions may differ. The PubMLST platform, which hosts many cgMLST schemes, provides versioned schemes and tools for comparing isolates within a defined scheme version. ## At a Glance: CgMLST Scheme Considerations by Organism | Organism | Scheme Scale | Key Implementation Consideration | Primary Surveillance Use | | --- | --- | --- | --- | | Salmonella enterica | Medium to large core genome schemes | Schema choice and software both affect cluster results, CRISPR-MVLST may offer higher discrimination for source tracking | Foodborne outbreak detection and source attribution | | Streptococcus pneumoniae | 1420 loci in the all-loci PubMLST scheme | Layered analysis with 7-gene MLST and rMLST supports continuity with historical data | Invasive disease surveillance and vaccine impact monitoring | | Neisseria gonorrhoeae | 1430 core genes in refined cgMLST | Horizontal gene transfer disrupts phylogenetic signals, LIN codes provide hierarchical lineage context | Antimicrobial resistance surveillance and lineage tracking | | Campylobacter jejuni | Schemas vary by up to twofold in locus number | Cluster assignment is strongly affected by schema choice | Foodborne outbreak detection | | Shiga toxin-producing Escherichia coli | Species-specific core genome schemes | Intra-species contamination is a major source of variability in cluster decisions | Foodborne outbreak detection and clinical surveillance | | Listeria monocytogenes | Species-specific core genome schemes | Read quality and assembly quality influence sample inclusion | Foodborne outbreak detection and regulatory response | ## Practical Workflow for CgMLST Implementation ### Step 1: Define the Surveillance Objective Before selecting a cgMLST scheme, clarify what the typing results will be used for. Outbreak detection requires high resolution to distinguish closely related isolates. Long-term population surveillance may prioritize stability and comparability with historical data. Antimicrobial resistance surveillance may require integration of resistance gene profiles with lineage information. The Salmonella field study demonstrated that different methods serve different purposes. Traditional serotyping identified nine serovars among 61 isolates, while cgMLST identified 18 types and CRISPR-MVLST identified 30 types. A laboratory focused on source tracking along a food chain might prefer a method with higher discrimination, while a laboratory maintaining long-term national surveillance might prioritize the comparability of cgMLST with historical MLST data. See the [Salmonella subtyping method comparison](https://pubmed.ncbi.nlm.nih.gov/36194132). ### Step 2: Select a Scheme and Software Platform Choose a cgMLST scheme that is maintained by a recognized reference database such as PubMLST or NCBI Pathogen Detection. The [NCBI Pathogen Detection resource](https://www.ncbi.nlm.nih.gov/pathogens) provides access to pathogen genomic data and analysis tools for foodborne pathogens and other organisms of public health importance. Consider whether the scheme is compatible with your sequencing platform and bioinformatics infrastructure. Some schemes are designed for assembly-based analysis, while others can work with raw reads. The inter-laboratory study found that cgMLST was mainly reproducible across different bioinformatics tools, but for Salmonella enterica, discrepancies due to different software were larger than discrepancies due to different schemas. For other species, different schemas introduced larger discrepancies than different software. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Step 3: Establish Quality Control Procedures Quality control begins with the sequencing data. The inter-laboratory study found that differences in the interpretation of raw sequence and genome assembly quality influenced sample inclusion and ultimately cluster composition. Intra-species contamination was the most significant factor driving variability in decisions on whether to include or exclude samples. Implement routine checks for: - Read quality scores and coverage depth - Assembly metrics such as contig count and N50 - Presence of expected species-specific marker genes - Absence of contamination from other species or from the same species in mixed samples The [Bacteriological Analytical Manual from the FDA](https://www.fda.gov/food/laboratory-methods-food/bacteriological-analytical-manual-bam) provides context for official food microbiology methods and quality expectations that can inform laboratory workflows. ### Step 4: Validate with Authenticated Control Strains Use authenticated biological materials to validate your cgMLST pipeline. The [American Type Culture Collection](https://www.atcc.org/) provides reference strains with documented provenance that can serve as positive controls for allele calling and cluster analysis. Run control strains through the entire workflow from DNA extraction to cluster assignment. Confirm that expected relationships are recovered and that allele calls match published values for the control strains. Document any discrepancies and investigate whether they reflect pipeline errors or genuine genetic variation. ### Step 5: Define Cluster Interpretation Thresholds Cluster thresholds should be established using validated data for your organism and scheme. The threshold represents the maximum allele distance within which isolates are considered part of the same cluster. Thresholds are not universal. They depend on the genetic diversity of the species, the number of loci in the scheme, and the intended use of the clusters. A threshold that works for outbreak detection may be too stringent for population surveillance. The inter-laboratory study showed that cluster assignment for Campylobacter jejuni was strongly affected by cgMLST schemas differing by a factor of two in the number of loci. This finding underscores the need to document the schema version and threshold for every cluster assignment. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Step 6: Document and Archive Analysis Parameters Maintain records of: - Scheme name and version - Software and version used for allele calling - Reference database version - Quality metrics for each isolate - Cluster threshold applied - Date of analysis and analyst identification These records support reproducibility and allow retrospective reanalysis when schemes are updated. ## Options and Tradeoffs in CgMLST Scheme Selection ### Assembly-Based versus Read-Based Approaches Some cgMLST pipelines require genome assembly before allele calling, while others can call alleles directly from raw reads. Assembly-based approaches may be more sensitive to assembly quality, while read-based approaches may be faster but less robust for repetitive regions. The Czech pneumococcal study used the Velvet de novo assembler to process Illumina MiSeq data before uploading assembled genomes to the PubMLST database for automated characterization. This assembly-based workflow is common for schemes hosted on the BIGSdb platform. See the [Czech Republic pneumococcal surveillance report](https://pubmed.ncbi.nlm.nih.gov/33086855). ### Single Scheme versus Multi-Resolution Analysis Some laboratories implement multiple resolution levels within a single workflow. The Czech pneumococcal approach compared isolates at three levels: seven MLST genes, 53 ribosomal genes, and 1420 genes. This layered strategy allows laboratories to maintain continuity with historical MLST data while gaining the resolution of cgMLST. The Neisseria gonorrhoeae LIN code system takes a similar multi-resolution approach by conveying lineage information at multiple levels within a single code. The LIN codes relate to familiar typing schemes such as Ng cgMLST v1, 7-locus MLST, and NG-STAR clonal complex, which facilitates communication between laboratories using different historical schemes. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). ### Public Databases versus Local Pipelines Public databases such as PubMLST and NCBI Pathogen Detection offer maintained schemes, curated allele databases, and web-based analysis tools. Local pipelines offer greater control over parameters and data handling but require bioinformatics expertise and ongoing maintenance. The [NCBI Pathogen Detection resource](https://www.ncbi.nlm.nih.gov/pathogens) provides integrated analysis for foodborne pathogens and other organisms, which can reduce the burden of local pipeline development. However, laboratories with specialized needs may find that local pipelines offer necessary flexibility. ## Records and Measurements for CgMLST Implementation ### Essential Records for Each Isolate Maintain a laboratory information system that captures: - Isolate identifier and source information - DNA extraction method and quality metrics - Sequencing platform and run identifier - Read quality metrics - Assembly metrics if assembly-based analysis is used - Allele calls for each locus in the scheme - Allele distance to reference isolates or cluster representatives - Cluster assignment and threshold used - Scheme version and database version ### Quality Metrics to Monitor Track the following metrics across runs to identify systematic issues: - Percentage of loci successfully called per isolate - Number of isolates failing quality thresholds - Reproducibility of allele calls for control strains - Turnaround time from sample receipt to cluster assignment - Frequency of schema updates and their impact on cluster assignments ### Common Failure Patterns in CgMLST Analysis | Failure Pattern | Likely Cause | Recommended Action | | --- | --- | --- | | Low percentage of loci called | Poor assembly quality or contamination | Review read quality, check for mixed cultures, re-extract if needed | | Inconsistent allele calls for control strains | Pipeline version changes or database updates | Validate pipeline after any software or database update | | Cluster assignments change between runs | Different schema versions or thresholds | Document all parameters and reanalyze historical data when schemes change | | Discrepancies between software platforms | Different allele calling algorithms | Select one primary platform and validate against reference data | | Intra-species contamination detected | Mixed cultures or cross-contamination during processing | Repeat isolation and purification before sequencing | ## Common Failure Patterns and How to Avoid Them ### Failure to Track Scheme Versions Laboratories that do not record which scheme version was used for each analysis cannot reliably compare results over time. Scheme updates may change locus composition, allele definitions, or both. When a scheme is updated, historical isolates may need to be reanalyzed to maintain comparability. ### Inconsistent Quality Thresholds The inter-laboratory study found that differences in the interpretation of raw sequence and genome assembly quality influenced sample inclusion and cluster composition. Laboratories that apply inconsistent quality thresholds may include or exclude samples differently across runs, which can change cluster assignments even when the underlying data are identical. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Overreliance on a Single Threshold A cluster threshold that is appropriate for one species or scheme may not transfer to another. Laboratories should validate thresholds using local isolate collections and reference data instead of adopting thresholds from published studies without verification. ### Ignoring Horizontal Gene Transfer Some bacterial species, particularly Neisseria gonorrhoeae, undergo extensive horizontal gene transfer that disrupts phylogenetic signals. Conventional typing systems may fail to reflect strain taxonomy reliably. The LIN code system was developed specifically to address this challenge by providing a hierarchical nomenclature that remains stable despite recombination. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). ## Biosafety and Laboratory Practice Considerations ### Risk Assessment for Bacterial Isolates Laboratories working with bacterial isolates for cgMLST analysis must conduct risk assessments based on the biosafety level appropriate for the organism. The [CDC and NIH Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html) provides principles for risk assessment, containment, and decontamination that apply to laboratory workflows. Risk assessment should consider: - The pathogenicity of the organism - The route of transmission - The concentration and volume of cultures - The procedures performed, including DNA extraction and sequencing library preparation - The potential for aerosol generation ### Decontamination and Waste Handling Standard laboratory decontamination procedures should be applied to cultures, spent media, and contaminated materials. The [Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html) guidance provides principles for decontamination that laboratories should incorporate into their standard operating procedures. ### Professional Training and Competency Laboratory personnel performing cgMLST analysis should have appropriate training in both molecular biology and bioinformatics. The [American Society for Microbiology](https://asm.org/) provides professional education resources and practice standards that support workforce development in clinical and public health microbiology. ## Antimicrobial Resistance Context for CgMLST ### Integrating Resistance Gene Profiles with Lineage Information CgMLST provides lineage information that can be correlated with antimicrobial resistance phenotypes and genotypes. The Neisseria gonorrhoeae LIN code study demonstrated that clustering with LIN codes provides insight into associations between genotype and phenotype for traits such as antibiotic resistance. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). The [World Health Organization antimicrobial resistance page](https://www.who.int/health-topics/antimicrobial-resistance) defines antimicrobial resistance as a global public health concern and emphasizes the importance of surveillance for tracking resistant organisms. CgMLST can support antimicrobial resistance surveillance by identifying clonal spread of resistant lineages. ### Antimicrobial Susceptibility Testing Context CgMLST does not replace phenotypic antimicrobial susceptibility testing. The [Clinical and Laboratory Standards Institute](https://clsi.org/standards/products/microbiology) provides standards for antimicrobial susceptibility testing that laboratories should follow for phenotypic confirmation of resistance. Genotypic prediction of resistance from whole genome sequence data is an active area of research, but it requires validated resistance gene databases and careful interpretation. CgMLST should be viewed as a complement to, not a replacement for, phenotypic susceptibility testing. ## Species-Specific Considerations ### Salmonella enterica The Salmonella field study along a poultry production chain provides a detailed example of cgMLST performance in a foodborne surveillance context. Among 61 isolates, cgMLST identified 18 types with a discrimination index of 0.8541. The study also demonstrated that CRISPR-MVLST detected 30 types with a higher discrimination index of 0.9628, suggesting that alternative methods may offer advantages for source tracking in some contexts. See the [Salmonella subtyping method comparison](https://pubmed.ncbi.nlm.nih.gov/36194132). For Salmonella, the inter-laboratory study found that discrepancies due to different software were larger than discrepancies due to different schemas. This finding suggests that laboratories using different software platforms may obtain different cluster results even when using the same schema. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Streptococcus pneumoniae The Czech Republic implementation of WGS for invasive pneumococcal disease surveillance provides a model for integrating cgMLST into national reference laboratory workflows. The laboratory compared isolates at three resolution levels and used the PubMLST database with the BIGSdb platform for automated characterization. See the [Czech Republic pneumococcal surveillance report](https://pubmed.ncbi.nlm.nih.gov/33086855). The pneumococcal all-loci scheme includes 1420 genes that cover MLST genes, ribosomal genes, and cgMLST genes. This comprehensive approach allows laboratories to move between resolution levels depending on the surveillance question. ### Neisseria gonorrhoeae Neisseria gonorrhoeae presents unique challenges for cgMLST because of extensive horizontal gene transfer. The LIN code system was developed using a refined 1430 core gene cgMLST and provides hierarchical lineage information within a single code. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). The LIN codes are automatically assigned and publicly accessible through the PubMLST Neisseria database. This system allows laboratories to relate new isolates to familiar typing schemes such as Ng cgMLST v1, 7-locus MLST, and NG-STAR clonal complex. ### Campylobacter jejuni Campylobacter jejuni cluster assignment is strongly affected by cgMLST schema choice. The inter-laboratory study found that schemas differing by a factor of two in the number of loci produced substantially different cluster assignments. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Laboratories working with Campylobacter should carefully document the schema version and validate cluster thresholds using local isolate collections. ### Shiga Toxin-Producing Escherichia coli The inter-laboratory study identified intra-species contamination as the most significant factor driving variability in decisions on whether to include or exclude samples for Shiga toxin-producing E. coli. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Laboratories should implement rigorous purity checks before sequencing and should investigate any samples with mixed allele calls at multiple loci. ### Listeria monocytogenes Listeria monocytogenes is a foodborne pathogen of regulatory concern, and cgMLST is widely used for outbreak detection and source attribution. The inter-laboratory study included L. monocytogenes in its assessment of cluster detection variability. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ## Implementation Steps for a New CgMLST Workflow ### Step 1: Assess Laboratory Infrastructure Evaluate existing sequencing capacity, computational resources, and bioinformatics expertise. Determine whether the laboratory can support assembly-based analysis or whether read-based approaches are more appropriate. ### Step 2: Select Reference Materials Acquire authenticated reference strains from the [American Type Culture Collection](https://www.atcc.org/) for pipeline validation. Select strains that represent the diversity of the species and that have published cgMLST allele calls. ### Step 3: Validate the Pipeline Run reference strains through the complete workflow from DNA extraction to cluster assignment. Compare allele calls with published values and investigate any discrepancies. ### Step 4: Establish Quality Control Procedures Define quality thresholds for read depth, assembly metrics, and locus calling rates. Document these thresholds in standard operating procedures. ### Step 5: Train Personnel Ensure that laboratory personnel understand the principles of cgMLST, the specific scheme being used, and the interpretation of cluster results. The [American Society for Microbiology](https://asm.org/) provides educational resources that can support training. ### Step 6: Pilot Testing Run a pilot set of isolates with known epidemiological relationships to confirm that the workflow produces expected cluster assignments. Document any discrepancies and refine procedures as needed. ### Step 7: Full Implementation Begin routine use of the cgMLST workflow, maintaining detailed records of all analysis parameters and quality metrics. ## Professional Escalation Criteria ### When to Escalate to a Reference Laboratory Contact a reference laboratory or scheme curator when: - A large proportion of isolates fail quality thresholds - Allele calls for control strains do not match published values - Cluster assignments are inconsistent across software platforms - A novel allele or locus pattern suggests a possible new variant - Results have public health implications that require confirmation ### When to Reanalyze Historical Data Reanalyze historical data when: - The scheme version is updated - The reference database is substantially revised - A cluster threshold is changed - A new software version is implemented ### When to Seek Curator Input Contact scheme curators when: - An isolate has alleles that do not match any known allele in the database - The core genome definition appears to exclude loci that are present in your isolates - You need guidance on cluster threshold selection for a new species ## Limitations of CgMLST ### Resolution Limits CgMLST may not provide sufficient resolution for all surveillance questions. The Salmonella study found that CRISPR-MVLST had higher discrimination than cgMLST for source tracking along a food chain. See the [Salmonella subtyping method comparison](https://pubmed.ncbi.nlm.nih.gov/36194132). ### Impact of Horizontal Gene Transfer For species with extensive horizontal gene transfer, such as Neisseria gonorrhoeae, cgMLST may not fully capture population structure. The LIN code system was developed to address this limitation. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). ### Schema Dependency Results are dependent on the schema version and software platform. The inter-laboratory study demonstrated that different schemas and software can produce different cluster assignments. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Quality Sensitivity CgMLST results are sensitive to sequencing and assembly quality. The inter-laboratory study found that differences in the interpretation of raw sequence and genome assembly quality influenced sample inclusion and cluster composition. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Database Maintenance Burden CgMLST schemes require ongoing maintenance as new alleles and isolates are discovered. Laboratories that rely on public databases must track updates and reanalyze data when necessary. ## Safety and Regulatory Context ### Laboratory Biosafety All work with bacterial isolates must follow appropriate biosafety practices. The [CDC and NIH Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html) provides guidance for risk assessment, containment, and decontamination. ### Food Safety Context For foodborne pathogens, cgMLST results may inform regulatory actions. The [FDA Bacteriological Analytical Manual](https://www.fda.gov/food/laboratory-methods-food/bacteriological-analytical-manual-bam) provides official methods context for food microbiology laboratories. ### Antimicrobial Resistance Surveillance CgMLST supports antimicrobial resistance surveillance by identifying clonal spread of resistant lineages. The [World Health Organization antimicrobial resistance page](https://www.who.int/health-topics/antimicrobial-resistance) emphasizes the importance of surveillance for addressing antimicrobial resistance as a global health threat. ## A Practical Decision Framework for Selecting and Validating cgMLST Schemes Selecting a cgMLST scheme involves more than choosing a database entry. Laboratories must weigh the surveillance objective, the genetic structure of the target species, the available bioinformatics capacity, and the need for comparability with historical data. This section provides a structured decision framework that laboratories can apply before committing to a scheme, along with a validation protocol and a troubleshooting method for resolving discrepancies that arise during routine use. ### Step 1: Define the Required Resolution Level The first decision point is the resolution level needed for the surveillance question. The Czech pneumococcal surveillance program demonstrated a layered approach that compared isolates at three levels: seven MLST genes, 53 ribosomal genes, and 1420 genes that included MLST genes, ribosomal genes, and cgMLST loci. This design allowed the laboratory to maintain continuity with historical MLST data while gaining the higher resolution of a full core genome comparison. See the [Czech Republic implementation report on pneumococcal WGS surveillance](https://pubmed.ncbi.nlm.nih.gov/33086855). Laboratories should ask whether the typing results will support outbreak detection, long-term population surveillance, or antimicrobial resistance monitoring. Outbreak detection typically requires the highest available resolution to distinguish closely related isolates. Long-term surveillance may prioritize stability and comparability with historical data. The Salmonella field study along a poultry production chain found that cgMLST identified 18 types among 61 isolates with a discrimination index of 0.8541, while CRISPR-MVLST detected 30 types with a discrimination index of 0.9628. A laboratory focused on source tracking might prefer the higher discrimination of CRISPR-MVLST, while a laboratory maintaining national surveillance might prioritize the comparability of cgMLST with historical MLST data. See the [Microbiology Spectrum study on subtyping methods for foodborne Salmonella surveillance](https://pubmed.ncbi.nlm.nih.gov/36194132). ### Step 2: Assess the Genetic Structure of the Target Species The genetic architecture of the species determines how much resolution a cgMLST scheme can provide. Species with extensive horizontal gene transfer, such as Neisseria gonorrhoeae, present challenges because recombination disrupts phylogenetic signals and can make conventional typing systems unreliable. The LIN code system was developed using a refined 1430 core gene cgMLST to address this challenge, providing hierarchical lineage information within a single code that relates to familiar typing schemes such as Ng cgMLST v1, 7-locus MLST, and NG-STAR clonal complex. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). For species with limited recombination, standard cgMLST schemes may be sufficient. The inter-laboratory study found that cgMLST was mainly reproducible across different bioinformatics tools for most species, but for Salmonella enterica, discrepancies due to different software were larger than discrepancies due to different schemas. For other species, different schemas introduced larger discrepancies than different software. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Laboratories should review the published literature for their target species to understand whether recombination is a known issue and whether alternative approaches such as LIN codes or SNP calling are recommended. ### Step 3: Evaluate Scheme Locus Composition and Version Stability Before adopting a scheme, examine the number of loci and the version history. The Campylobacter jejuni example is instructive: different cgMLST schemas can differ by a factor of two in the number of loci, and cluster assignment is strongly affected by this difference. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Laboratories should document the following for any candidate scheme: - The total number of loci in the scheme - The version number and release date - The number of alleles defined for each locus - The frequency of version updates - The criteria used to define core genome loci - The reference strains used to validate the scheme Schemes with frequent version updates may require more frequent reanalysis of historical data. Schemes with a small number of loci may not provide sufficient resolution for outbreak detection. Schemes with a very large number of loci may be more sensitive to assembly quality and missing data. ### Step 4: Compare Software Platform Compatibility The inter-laboratory study demonstrated that software choice can affect cluster results. For Salmonella enterica, discrepancies due to different software were larger than discrepancies due to different schemas. For other species, different schemas introduced larger discrepancies than different software. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Laboratories should test candidate schemes on their existing bioinformatics infrastructure before committing to routine use. This testing should include: - Running a set of reference isolates through the complete workflow - Comparing allele calls across software platforms if multiple platforms are available - Assessing whether the software supports the required scheme version - Evaluating the time and computational resources required for analysis The Czech pneumococcal study used the Velvet de novo assembler to process Illumina MiSeq data before uploading assembled genomes to the PubMLST database using the BIGSdb platform. This assembly-based workflow is common for schemes hosted on PubMLST. See the [Czech Republic pneumococcal surveillance report](https://pubmed.ncbi.nlm.nih.gov/33086855). ### Step 5: Establish a Validation Protocol with Authenticated Control Strains Validation is a critical step that laboratories sometimes skip or perform inconsistently. The [American Type Culture Collection](https://www.atcc.org/) provides authenticated reference strains with documented provenance that can serve as positive controls for allele calling and cluster analysis. A validation protocol should include the following elements: - Select at least three reference strains that represent the genetic diversity of the species - Confirm that the reference strains have published cgMLST allele calls or sequence types - Run the reference strains through the complete workflow from DNA extraction to cluster assignment - Compare allele calls with published values - Confirm that expected relationships between reference strains are recovered - Document any discrepancies and investigate whether they reflect pipeline errors or genuine genetic variation The validation protocol should be repeated whenever the software, scheme version, or reference database is updated. ### Step 6: Define Cluster Thresholds Using Local Isolate Collections Cluster thresholds should not be adopted from published studies without verification. The appropriate threshold depends on the genetic diversity of the local isolate collection, the number of loci in the scheme, and the intended use of the clusters. A practical approach for defining thresholds is to: - Collect a set of isolates with known epidemiological relationships, such as isolates from a confirmed outbreak - Calculate the pairwise allele distances between these isolates - Determine the maximum allele distance within the outbreak cluster - Compare this distance with the distances between unrelated isolates - Select a threshold that separates outbreak-related isolates from unrelated isolates The inter-laboratory study found that differences in the interpretation of raw sequence and genome assembly quality influenced sample inclusion and cluster composition. Intra-species contamination was the most significant factor driving variability in decisions on whether to include or exclude samples. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Step 7: Document the Decision and Archive the Rationale The final step in the decision framework is documentation. Laboratories should maintain a written record of: - The surveillance objective that drove the scheme selection - The candidate schemes evaluated and the reasons for rejection - The validation results for the selected scheme - The cluster threshold and the evidence supporting it - The software platform and version used - The date of the decision and the personnel involved This documentation supports reproducibility and provides a basis for revisiting the decision when new information becomes available. ### A Record System for Tracking Scheme Performance Once a scheme is implemented, laboratories should track its performance over time. A simple record system can capture the following metrics for each analysis run: | Metric | Purpose | Action When Out of Range | | --- | --- | --- | | Percentage of loci successfully called | Detects assembly or contamination problems | Review read quality, check for mixed cultures | | Number of isolates failing quality thresholds | Identifies systematic issues | Investigate extraction or sequencing protocols | | Reproducibility of control strain allele calls | Detects pipeline or database changes | Validate pipeline after any update | | Turnaround time from sample receipt to cluster assignment | Monitors workflow efficiency | Review bottlenecks in the analysis pipeline | | Frequency of schema updates | Anticipates reanalysis needs | Plan for reanalysis when updates occur | The inter-laboratory study found that cgMLST was mainly reproducible across different bioinformatics tools, but the largest influence on cgMLST-defined clusters was the inclusion or exclusion of samples. This finding underscores the importance of consistent quality thresholds and documentation. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### Troubleshooting Method for Cluster Assignment Discrepancies When cluster assignments differ between runs, between software platforms, or between laboratories, a systematic troubleshooting method can identify the source of the discrepancy. The following sequence addresses the most common causes: #### Step 1: Verify Sample Identity and Purity Intra-species contamination was the most significant factor driving variability in decisions on whether to include or exclude samples in the inter-laboratory study. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). Check the original culture for purity by streaking for isolated colonies and confirming colony morphology. Review the sequencing data for evidence of mixed allele calls at multiple loci. If contamination is suspected, repeat isolation and purification before resequencing. #### Step 2: Confirm Scheme Version and Database Version Compare the scheme version and reference database version used in each analysis. Differences in version can change locus composition and allele definitions, which will change cluster assignments. Reanalyze the data using the same scheme version and database version to confirm that the discrepancy resolves. #### Step 3: Compare Quality Metrics Review the read quality, assembly metrics, and locus calling rates for the discrepant isolates. The inter-laboratory study found that differences in the interpretation of raw sequence and genome assembly quality influenced sample inclusion and cluster composition. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). If one analysis used lower-quality data, the allele calls may be incomplete or incorrect. Reanalyze the data after applying consistent quality thresholds. #### Step 4: Evaluate Software Platform Differences For Salmonella enterica, discrepancies due to different software were larger than discrepancies due to different schemas. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). If different software platforms were used, compare the allele calls for the discrepant isolates across platforms. Identify the specific loci where the calls differ and investigate whether the difference reflects a genuine biological difference or a bioinformatics artifact. #### Step 5: Assess Threshold Sensitivity Cluster assignments can change when the threshold is near the boundary of the observed allele distances. Calculate the pairwise allele distances for the discrepant isolates and determine whether the distances are close to the threshold. If so, the discrepancy may reflect threshold sensitivity instead of a data problem. #### Step 6: Escalate to the Scheme Curator If the discrepancy persists after completing the previous steps, contact the scheme curator for guidance. Provide the isolate identifiers, the scheme version, the software platform, and the quality metrics for the discrepant isolates. The curator can determine whether the discrepancy reflects a novel allele, a database error, or a known issue with the scheme. ### Common Failure Patterns in Scheme Implementation | Failure Pattern | Likely Cause | Recommended Action | | --- | --- | --- | | Low percentage of loci called | Poor assembly quality or contamination | Review read quality, check for mixed cultures, re-extract if needed | | Inconsistent allele calls for control strains | Pipeline version changes or database updates | Validate pipeline after any software or database update | | Cluster assignments change between runs | Different scheme versions or thresholds | Document all parameters and reanalyze historical data when schemes change | | Discrepancies between software platforms | Different allele calling algorithms | Select one primary platform and validate against reference data | | Intra-species contamination detected | Mixed cultures or cross-contamination during processing | Repeat isolation and purification before sequencing | | Novel alleles not recognized by the database | Genuine genetic variation or database gap | Contact the scheme curator for guidance | ### Professional Escalation Criteria for Scheme Selection and Validation Laboratories should escalate to a reference laboratory or scheme curator when: - The validation protocol fails to reproduce published allele calls for authenticated control strains - A large proportion of isolates have low locus calling rates despite acceptable read quality - Cluster assignments are inconsistent across software platforms for the same scheme version - A novel allele pattern suggests a possible new variant that may require database updates - The laboratory lacks the bioinformatics expertise to resolve discrepancies between software platforms The [American Society for Microbiology](https://asm.org/) provides professional education resources that can support workforce development in clinical and public health microbiology, including training in whole genome sequence analysis. ### Biosafety Considerations for Validation Work Validation work involves handling authenticated reference strains and clinical or environmental isolates. Laboratories must conduct risk assessments based on the biosafety level appropriate for the organism. The [CDC and NIH Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html) provides principles for risk assessment, containment, and decontamination that apply to laboratory workflows. Risk assessment should consider the pathogenicity of the organism, the route of transmission, the concentration and volume of cultures, the procedures performed including DNA extraction and sequencing library preparation, and the potential for aerosol generation. Standard laboratory decontamination procedures should be applied to cultures, spent media, and contaminated materials. ### Antimicrobial Resistance Surveillance Context The decision framework should also consider whether the cgMLST results will be integrated with antimicrobial resistance surveillance. The Neisseria gonorrhoeae LIN code study demonstrated that clustering with LIN codes provides insight into associations between genotype and phenotype for traits such as antibiotic resistance. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). The [World Health Organization antimicrobial resistance page](https://www.who.int/health-topics/antimicrobial-resistance) defines antimicrobial resistance as a global public health concern and emphasizes the importance of surveillance for tracking resistant organisms. CgMLST can support antimicrobial resistance surveillance by identifying clonal spread of resistant lineages, but it does not replace phenotypic antimicrobial susceptibility testing. The [Clinical and Laboratory Standards Institute](https://clsi.org/standards/products/microbiology) provides standards for antimicrobial susceptibility testing that laboratories should follow for phenotypic confirmation of resistance. ### Food Safety Regulatory Context For foodborne pathogens, cgMLST results may inform regulatory actions. The [FDA Bacteriological Analytical Manual](https://www.fda.gov/food/laboratory-methods-food/bacteriological-analytical-manual-bam) provides official methods context for food microbiology laboratories. Laboratories that generate cgMLST data for regulatory purposes should ensure that their workflows meet the quality expectations described in official methods and that their records support audit and review. The [NCBI Pathogen Detection resource](https://www.ncbi.nlm.nih.gov/pathogens) provides access to pathogen genomic data and analysis tools for foodborne pathogens and other organisms of public health importance. Laboratories may use this resource to compare their isolates with publicly available data and to participate in broader surveillance efforts. ## Frequently Asked Questions ### What is the difference between MLST and cgMLST? Classical MLST compares seven housekeeping genes and assigns a sequence type. CgMLST compares hundreds or thousands of core genes and provides higher resolution for distinguishing closely related isolates. The Czech pneumococcal study compared isolates at three levels including seven MLST genes, 53 ribosomal genes, and 1420 genes, demonstrating how cgMLST extends classical MLST resolution. See the [Czech Republic pneumococcal surveillance report](https://pubmed.ncbi.nlm.nih.gov/33086855). ### How do I choose a cgMLST scheme for my organism? Select a scheme that is maintained by a recognized reference database such as PubMLST or NCBI Pathogen Detection. Consider the number of loci, the species diversity, and the intended surveillance purpose. The inter-laboratory study found that schema choice had different impacts depending on the species, with Campylobacter jejuni being strongly affected by schema differences. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### What quality metrics should I monitor for cgMLST? Monitor read quality, assembly metrics, percentage of loci successfully called, and reproducibility of control strain results. The inter-laboratory study found that intra-species contamination was the most significant factor driving variability in sample inclusion decisions. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ### How do I determine a cluster threshold for cgMLST? Cluster thresholds should be validated using local isolate collections and reference data. Thresholds are species-specific and depend on the number of loci in the scheme. The Salmonella study used cgMLST as the gold standard and identified 18 types among 61 isolates, but the appropriate threshold depends on the surveillance question. See the [Salmonella subtyping method comparison](https://pubmed.ncbi.nlm.nih.gov/36194132). ### Can cgMLST predict antimicrobial resistance? CgMLST provides lineage information that can be correlated with resistance phenotypes and genotypes, but it does not replace phenotypic susceptibility testing. The Neisseria gonorrhoeae LIN code study demonstrated associations between genotype and phenotype for antibiotic resistance traits. See the [Neisseria gonorrhoeae LIN code study](https://doi.org/10.7554/elife.107758). Follow [CLSI standards for antimicrobial susceptibility testing](https://clsi.org/standards/products/microbiology) for phenotypic confirmation. ### What should I do if my isolates do not match any known alleles? Contact the scheme curator for guidance. Novel alleles may indicate new variants that require addition to the reference database. Document the isolate information and the loci with novel alleles for further investigation. ### How often should I update my cgMLST scheme? Update the scheme when the curator releases a new version, but document the version used for each analysis. When schemes are updated, historical isolates may need to be reanalyzed to maintain comparability. Track scheme versions in your laboratory information system. ### What are the limitations of cgMLST for outbreak detection? CgMLST may not provide sufficient resolution for all outbreak investigations. The Salmonella study found that CRISPR-MVLST had higher discrimination than cgMLST for source tracking. See the [Salmonella subtyping method comparison](https://pubmed.ncbi.nlm.nih.gov/36194132). Results are also sensitive to schema choice and bioinformatics pipeline, as demonstrated in the inter-laboratory study. See the [inter-laboratory study on cluster detection variability](https://doi.org/10.3389/fmicb.2025.1629731). ## Related Bacteriology Guides - [Bacterial Carriage in Ducks: Pathogens of Public Health Concern](/knowledge/bacteria/avian-bacteria/bacteria-carried-by-ducks) - [Bacterial Contamination of Chicken Meat: Food Safety and Public Health](/knowledge/bacteria/avian-bacteria/bacterial-contamination-chicken-meat) - [Bacterial Pathogens in Chicken Meat: Risk Assessment and Public Health Impact](/knowledge/bacteria/avian-bacteria/bacterial-pathogens-chicken-meat-risk-assessment-public-health) - [Salmonellosis and Colibacillosis in Poultry: Bacterial Pathogens, Public Health Implications, and Control Strategies](/knowledge/bacteria/avian-bacteria/salmonellosis-colibacillosis-poultry-bacterial-pathogens) - [Bacterial Foodborne Pathogens in Poultry: Salmonella and Escherichia coli in Chickens, Clinical and Public Health Perspectives](/knowledge/bacteria/avian-bacteria/bacterial-foodborne-pathogens-poultry-salmonella-escherichia-coli-chickens) ## References and Further Reading - [Biosafety in Microbiological and Biomedical Laboratories](https://www.cdc.gov/labs/bmbl/index.html). CDC and NIH. - [Microbiology Resources](https://asm.org/). American Society for Microbiology. - [Antimicrobial Resistance](https://www.who.int/health-topics/antimicrobial-resistance). World Health Organization. - [Bacteriological Analytical Manual](https://www.fda.gov/food/laboratory-methods-food/bacteriological-analytical-manual-bam). U.S. Food and Drug Administration. - [Antimicrobial Susceptibility Testing](https://clsi.org/standards/products/microbiology). Clinical and Laboratory Standards Institute. - [Culture Collections](https://www.atcc.org/). American Type Culture Collection. - [NCBI Pathogen Detection](https://www.ncbi.nlm.nih.gov/pathogens). National Center for Biotechnology Information. - [Comprehensive Assessment of Subtyping Methods for Improved Surveillance of Foodborne Salmonella.](https://pubmed.ncbi.nlm.nih.gov/36194132). Microbiology spectrum, 2022. - [Implementation and use of whole genome sequencing (WGS) in the surveillance of invasive pneumococcal disease, Czech Republic, 2017-2019.](https://pubmed.ncbi.nlm.nih.gov/33086855). Epidemiologie, mikrobiologie, imunologie : casopis Spolecnosti pro epidemiologii a mikrobiologii Ceske lekarske spolecnosti J.E. Purkyne, 2020. - [<,i>,Neisseria gonorrhoeae<,/i>, LIN codes provide a robust, multi-resolution lineage nomenclature.](https://doi.org/10.7554/elife.107758). 2025. - [An inter-laboratory study characterizes the impact of bioinformatic approaches on genome-based cluster detection for foodborne bacterial pathogens.](https://doi.org/10.3389/fmicb.2025.1629731). 2025. > This article is educational and does not replace validated laboratory procedures, clinical diagnosis, institutional biosafety review, or public-health guidance.