ctDNA vs MRD Testing: Understanding the Next Generation of Cancer Monitoring

For commercial breeding programmes, the discussion around SNP array vs genotyping by sequencing has evolved considerably. The challenge is no longer generating more markers. Most breeding organisations already have access to vast quantities of genomic data. The real challenge is generating the right information quickly enough to support advancement decisions, marker-assisted selection strategies, genomic selection pipelines and breeding programme optimisation.

Whether supporting marker-assisted selection, genomic selection, parentage QC, trait introgression, or other targeted breeding applications, organisations increasingly require genotyping workflows that can evolve alongside breeding objectives. New trait markers emerge, genomic prediction models improve and breeding priorities shift. Under these conditions, flexibility can be just as important as throughput and enables futureproofing genotyping screening strategies.

When evaluating Amplicon Sequencing, microarray vs NGS technologies or broader SNP array vs sequencing strategies, breeders should consider not only assay costs but also how effectively a platform supports long-term programme efficiency, breeding-cycle timelines and genetic gain.

Why cost per sample is the wrong number

Breeding programmes do not genotype material because they need more data. They genotype material because they need to make better decisions.

This distinction is often overlooked when comparing SNP array vs genotyping by sequencing platforms.

Cost per sample remains a useful procurement metric, but it rarely reflects the true value delivered by a genotyping workflow. More meaningful measures include:

• Cost per informative marker
• Cost per successful genotype call
• Cost per actionable data point
• Turnaround time to selection
• Call rate across breeding populations
• Ability to deploy new trait markers
• Impact on breeding-cycle efficiency 
• Increasing genetic gain

Which platform is cheaper?

For programmes analysing hundreds or a few thousand markers, targeted genotyping by sequencing often delivers a lower cost per informative data point because sequencing capacity is focused on loci that directly influence breeding decisions. However, SNP arrays can remain highly-cost effective where fixed, high-density marker sets are routinely applied across stable populations, particularly when array development and validation costs have already been amortised over large sample volumes and multiple breeding cycles.

In practice, the most economical platform depends on how marker data are used within a breeding programme. If selection decisions rely on only a subset of markers, generating tens of thousands of additional data points may provide limited operational value. At the same time, breeding objectives, trait targets and genomic prediction models continue to evolve. Under these conditions, technologies that enable marker panels to be updated efficiently can offer additional long-term value. As a result, many breeding organisations evaluate platforms not only on cost per sample, but also on marker utilisation, assay flexibility and their ability to support future breeding requirements.

How SNP arrays work, and where the cost floor sits

SNP arrays are a specialised type of microarray technology that genotype predefined markers through probe hybridisation and fluorescence intensity measurements. 

Array performance depends on: 

• Probe design quality 
• Hybridisation efficiency 
• Cluster separation 
• Signal intensity distributions 
• Genotype-calling algorithms 
• Marker validation during assay development 

For established plant and animal breeding programmes, SNP arrays provide: 

• Mature high-throughput agricultural genomics workflows 
• Consistent marker content 
• Well-established bioinformatics pipelines 
• High call rates for validated markers 
• Compatibility with historical datasets
• Proven support for genomic selection 

However, SNP arrays operate using a fixed panel architecture. When new disease-resistance loci, quality traits or genomic selection targets emerge, introducing those markers often requires redesigning and manufacturing new array content. In addition, as reference genomes improve and genome annotations become more refined, the positional context and biological interpretation of previously selected markers may change, potentially affecting the long-term relevance of fixed marker sets. 

For breeding programmes with rapidly evolving trait portfolios, this can create both financial and operational constraints. 

Another consideration is ascertainment bias. Many arrays were developed using elite breeding material. When applied to landraces, donor germplasm, wild relatives or introgression populations, marker informativeness can decline because variants present within those populations were not included during panel development. This effect may be amplified when arrays developed for one breeding population or geographic region are deployed more broadly across diverse global germplasm, where allele frequencies and haplotype structures can differ substantially. Studies comparing SNP arrays and sequencing technologies have demonstrated differences in marker representation, haplotype coverage and diversity analyses across different germplasm sets. [1] 

How targeted amplicon sequencing works 

Amplicon Sequencing is a form of targeted genotyping by sequencing that combines multiplex PCR, sample indexing and next-generation sequencing (NGS) to analyse selected genomic regions. 

Unlike microarray-based genotyping, which infers genotypes from hybridisation signals, Amplicon Sequencing generates genotype calls directly from sequencing reads using: 

• Read depth 
• Allele balance 
• Base-quality metrics 
• Locus recovery rates 
• Sequencing quality thresholds 

Modern Amp-Seq plant breeding workflows routinely multiplex hundreds to thousands of loci while simultaneously processing thousands of samples. Regardless of platform, successful genotyping begins with high-quality DNA extraction and sample preparation to ensure consistent marker recovery and reliable genotype calls. 

From a breeder's perspective, the most significant advantage is not simply the use of NGS. It is the ability to focus sequencing capacity on markers that directly influence breeding outcomes. 

Platforms such as Amp-Seq One and Flex-Seq™ support: 

• Marker-assisted selection 
• Genomic selection 
• Parentage QC 
• Germplasm characterisation 
• Trait-marker deployment 
• Molecular breeding workflows 
• Haplotype screening to reduce number of makers and increase cost efficiency 

As with any multiplex PCR workflow, panel performance depends on minimising locus dropout, maintaining uniform coverage and ensuring balanced amplification across targets. When optimised correctly, Amplicon Sequencing combines high throughput with the flexibility to adapt marker content as programme objectives evolve. 

This flexibility is particularly valuable when breeders need to deploy newly discovered markers rapidly, rather than waiting for the development of new fixed-array content. 

The broad adoption of genotyping-by-sequencing approaches has demonstrated their suitability for large-scale breeding programmes and high-diversity populations. Robert J. Elshire and colleagues showed that GBS workflows can deliver scalable and reproducible genotyping across genetically diverse species. [2] 

Side-by-side comparison: Density, flexibility, turnaround and cost per data point

Worked example: 5,000 samples, 500 markers

Consider a commercial plant genotyping programme in wheat processing 5,000 samples for:

• Genomic selection updates
• Disease-resistance screening
• Parentage QC
• Marker-assisted selection

The programme requires 500 validated markers to support advancement decisions.

If only 500 of 25,000 generated markers contribute to selection decisions, approximately 2% of generated marker content is directly informing operational outcomes. This does not imply that high-density arrays lack value. Rather, it highlights why breeders increasingly evaluate cost per informative marker and cost per actionable data point. 

The key question is not how many markers are generated. It is how many markers contribute to better breeding decisions. 

Turnaround time and breeding-cycle economics

Turnaround time remains one of the most overlooked components of genotyping economics.

Most breeders have experienced situations where the value of a genotype result was determined less by its cost and more by when it was delivered. 

Missing a crossing window, nursery advancement decision or planting deadline can result in undesirable material being carried through an additional breeding cycle. The economic consequences of these delays frequently exceed modest differences in assay pricing. 

For this reason, breeding programmes increasingly evaluate: 

• Time to data delivery 
• Seasonal throughput capacity 
• Laboratory scalability 
• Sample batching requirements 
• Flexibility to deploy new markers 
• Compatibility with breeding-cycle field costs 

In commercial breeding, delayed information can be more expensive than expensive information.

Ascertainment bias and diverse germplasm

Many SNP array vs microarray comparisons overlook ascertainment bias despite its importance in modern breeding programmes.

Although the SNP array vs microarray distinction is often discussed at the platform level, the more important consideration for breeders is whether marker content remains representative of their target germplasm.

As breeders increasingly incorporate: 

• Landraces 
• Wild relatives 
• Donor germplasm 
• Introgression populations 
• Pre-breeding materials 
• CRiSPR-edited varieties 

they often encounter genetic variation that may not be adequately represented in historical SNP-array designs. 

This becomes particularly important for programmes focused on climate resilience, disease resistance and broadening the genetic base of elite breeding material. 

Research comparing SNP arrays and genotyping-by-sequencing approaches has demonstrated that marker discovery methods influence haplotype tagging, marker representation and analytical outcomes. [1] 

For diverse germplasm sets, custom targeted sequencing panels can offer an advantage because marker content can be developed specifically around the populations being analysed. 

Imputation, low-pass whole-genome sequencing and marker density

Low-pass whole-genome sequencing, SNP arrays and targeted genotyping by sequencing each serve different purposes within agrigenomics workflows. 

Low-pass whole-genome sequencing provides broad genomic coverage but commonly relies on imputation to infer missing genotype calls. 

Similarly, lower-density SNP genotyping workflows may depend on reference populations to support genomic selection and downstream analyses. 

Imputation performs best when: 

• High-quality reference populations exist 
• Linkage disequilibrium is well characterised 
• Population structure is stable 
• Extensive genomic resources are available 

Population structure can substantially influence genomic selection accuracy and the effectiveness of genomic prediction models, particularly when training and validation populations differ, highlighting the importance of appropriate reference populations and study design. [4] 

However, these conditions may not be present for orphan crops, emerging breeding programmes or highly diverse germplasm landraces. 

Research by Davoud Torkamaneh and François Belzile demonstrated how imputation can enhance both GBS- and SNP-array-derived datasets while emphasising the importance of robust reference populations. [3] 

When a microarray is still the right choice 

Despite the growth of Amplicon Sequencing, SNP arrays remain an excellent solution for many applications. 

 Arrays are often appropriate when: 

• Marker content remains stable over multiple breeding cycles 
• Historical compatibility is essential 
• Very high marker density is required 
• Existing genomic selection models are array based 
• Large reference populations already exist 

For these programmes, the maturity and standardisation of microarray workflows can outweigh the advantages of increased panel flexibility. 

The goal is not to identify a universally superior technology. It is to identify the technology that best supports programme objectives. 

Choosing a platform for your breeding programme 

There is no universal winner in the SNP array vs sequencing debate because breeding programmes face different biological, operational and commercial challenges. 

SNP arrays continue to provide a robust solution for programmes built around stable marker content, extensive historical datasets and mature genomic selection frameworks. 

However, many breeding organisations operate in a very different environment than they did a decade ago. Trait portfolios evolve more rapidly, genomic prediction models are updated constantly and germplasm collections are becoming increasingly diverse. When trait-marker portfolios evolve from one breeding cycle to the next, the ability to update marker content quickly can become a competitive advantage. 

Amp-Seq One and Flex-Seq™ may be particularly valuable when: 

• Marker requirements evolve regularly 
• New trait markers must be deployed rapidly 
• Breeding cycles are time sensitive 
• Parentage QC is routinely performed 
• Diverse germplasm is evaluated 
• Custom marker panels are preferred 

SNP arrays may remain preferable when: 

• Marker content is fixed 
• Historical continuity is critical 
• Existing infrastructure is array based 
• Ultra-high marker densities are routinely required 

The critical question is no longer whether a platform can generate sufficient markers. Most breeding programmes already have access to more marker data than they routinely use. 

The challenge is ensuring that genotyping workflows remain aligned with evolving trait portfolios, genomic prediction models and breeding timelines. In that environment, both targeted genotyping by sequencing and SNP arrays play important roles, but the most effective solution is the one that delivers reliable, actionable information quickly enough to improve breeding decisions and accelerate genetic gain. 

References

1. Negro SS, Millet EJ, Madur D, Bauland C, Combes V, Welcker C, Tardieu F, Charcosset A and Nicolas SD. (2019). Genotyping-by-Sequencing and SNP-Arrays Are Complementary for Detecting Quantitative Trait Loci by Tagging Different Haplotypes in Association Studies. BMC Plant Biology, 19, 318. https://doi.org/10.1186/s12870-019-1926-4
2. Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES and Mitchell SE. (2011). A Robust, Simple Genotyping-by-Sequencing (GBS) Approach for High Diversity Species. PLOS ONE, 6(5), e19379. https://doi.org/10.1371/journal.pone.0019379
3. Torkamaneh D and Belzile F. (2015). Scanning and Filling: Ultra-Dense SNP Genotyping Combining Genotyping-by-Sequencing, SNP Array and Whole-Genome Resequencing Data. PLOS ONE, 10(7), e0131533. https://doi.org/10.1371/journal.pone.0131533
4. Werner CR, Gaynor RC, Gorjanc G, Hickey JM, Kox T, Abbadi A, Leckband G, Snowdon RJ and Stahl A. (2020). How Population Structure Impacts Genomic Selection Accuracy in Cross-Validation: Implications for Practical Breeding. Frontiers in Plant Science, 11, 592977. https://doi.org/10.3389/fpls.2020.592977