Application of SNV Detection Methods for Market Control of Food Products from New Genomic Techniques

Abstract

The detection of single-nucleotide variants (SNVs) is an important challenge in modern genomics, with broad applications in medicine, diagnostics, and agricultural biotechnology. Current detection approaches include PCR-based techniques with high-affinity probes, ligase-based strategies, and sequencing approaches, each with varying degrees of sensitivity, specificity, and practicality. Despite advances in SNV analysis in the medical field, their implementation in the official control and monitoring of genetically modified organisms (GMOs) remains limited. This challenge has gained priority with the advent of new genomic techniques (NGTs), such as CRISPR-Cas nucleases, which allow precise genome editing, including subtle changes at the nucleotide level without introducing foreign DNA. Therefore, traditional methods of GMO detection targeting transgene sequences may not be sufficient to monitor such GMOs. In the European Union, GMO legislation requires distinguishing between conventionally bred and genetically modified plants. The planned introduction of new regulatory categories of NGT plants (NGT1 and NGT2) with different surveillance requirements emphasizes the need for robust, sensitive, and cost-effective SNV detection methods suitable for distinguishing between GMOs, particularly in the context of food and feed safety, traceability, and compliance.

1. Introduction

The detection of SNVs is a well-established analytical approach widely used in medicine, diagnostics and agrobiotechnology [1,2,3]. SNVs, together with insertions and deletions, account for the majority of genetic variation between individuals. Besides SNV, the term SNP (single-nucleotide polymorphism) is also commonly used. A “variant” refers to any single-nucleotide change, regardless of its origin or frequency of occurrence. An SNP, on the other hand, refers to the germline (meaning it is heritable) and must be present in at least 1% of the population. Nevertheless, from a detection perspective, these changes are equivalent [4,5,6,7]. Therefore, in this review, all single-nucleotide substitutions are collectively referred to as SNVs.

A broad range of SNV detection methods have been developed, including PCR-based assays with high-affinity probes, ligation-based strategies, and sequencing approaches [8,9,10,11,12]. These techniques differ in sensitivity, specificity, cost, and complexity, and while they are routinely applied in medical diagnostics, their use outside the field remains limited. In plant breeding, SNV detection is mainly employed for marker-assisted selection, and, until recently, there was little need to adapt such methods for official GMO control [13,14,15,16,17,18,19].

Current EU GMO legislation is based on the detection of exogenous DNA elements introduced by transgenesis and requires analytical methods to meet defined minimum performance requirements (MPR), including robustness, specificity, and limit of detection [15,20]. The current strategy for the detection and identification of specific events involves performing screening, followed by event-specific analysis. The preliminary test seeks commonly used foreign genetic elements (e.g., the 35S promoter delivered from the cauliflower mosaic virus). There are 27 DNA screening elements that enable the detection of both authorized GMOs (currently 104 events for which specific methods are available) and unauthorized GMOs (for which the GM event-specific methods are unavailable) [21,22,23,24,25]. However, plants obtained through genome editing often only contain small nucleotide-level changes without foreign DNA. This challenges conventional GMO detection strategies and highlights the need for sensitive SNV-based methods that are suitable for market control of food and feed.

The proposed regulatory framework identifies two categories of NGT plants. NGT category 1 (NGT1) plants may contain up to 20 genetic modifications with a maximum size of 20 nucleotides each (excluding deletions which are not size restricted). Plants in this category are considered equivalent to conventionally bred; thus, the restrictions applied to traditional GMOs will not apply to them, but labeling their reproductive material as an NGT1 product is still planned. On the other hand, NGT category 2 (NGT2) are plants that do not meet NGT1 criteria, and the same restrictions will be applied to them as to conventional GMOs (with an easier authorization procedure in case of traits, with a positive impact on a sustainable agrifood system) [20,23,26,27,28]. As a result, we can predominantly expect gene edited plants with only minor changes to be placed on the market.

Oversight of genome editing in food and feed is governed by the European Food Safety Authority (EFSA) [22,29,30,31,32], within the framework of EU legislation, including Regulation (EC) No 1829/2003 [22] establishing authorization and labeling requirements, and Regulation (EU) No 619/2011 [25] defining analytical procedures and thresholds for unauthorized GM material.

This review aims to critically evaluate current SNV detection methods with respect to their applicability in official GMO and NGT control. The spotlight is placed on analytical performance parameters required by EU legislation (limit of detection (LOD), limit of quantification (LOQ), specificity, robustness, and quantification capability), and on identifying methodological gaps that limit the implementation of SNV-based approaches for market surveillance of NGT1 and NGT2 plants [20,23,26].

2. State of the Art in Point Mutation Detection

2.1. Methods Based on Target Sequence Amplification

For many years, SNV assays have been used in crop plants for variety identification and for detecting genetic markers associated with desirable traits. These include restriction fragment length polymorphism (RFLP), random amplified polymorphic DNA (RAPD), real-time PCR, and amplified fragment length polymorphism (AFLP), which were primarily applied for qualitative analyses or zygosity determination in the 0–0.5–1 system [13,14,16,17,18,19]. Historically, there has been no demand for quantitative SNV analyses in either GMOs or plants in general.

In contrast, numerous PCR-based SNV detection methods have been developed in medical diagnostics. Many of them simply rely on target amplification, but polymerase tolerance to primer–template mismatches can result in false positives [33,34,35]. To improve specificity and reliability, alternative approaches based on sequence-specific probes, ligation assays, or CRISPR/Cas systems have been developed [33,36,37,38,39].

2.1.1. Applications of Specific Probes

LNA Probes

Single-nucleotide variant detection frequently relies on oligonucleotides composed of nucleic acid analogues, used either as PCR probes/primers or as polymerase blockers [40]. Among these, locked nucleic acid (LNA) probes are the most widely applied due to their enhanced mismatch discrimination. The presence of a methylene bridge (2′-O,4′-C) increases duplex stability and binding energy, resulting in a markedly higher melting temperature difference between perfectly matched and mismatched hybrids, which may reach 20–30 °C for a single base mismatch [40].

As a result, LNA probes provide a relatively simple, cost-effective, and highly specific approach for SNV detection using PCR-based techniques, while also enabling quantitative analysis. Optimal performance depends on probe design, including oligonucleotide length (18–24 bases) and the central placement of LNA residues, with approximately half of the bases substituted by LNA [41]. This design enables the detection of DNA variants at ratios as low as 0.003% [42].

LNA oligonucleotides can also be applied to selectively suppress amplification of wild-type alleles (Figure 1) [40]. Novel LNA-based approaches continue to be developed, such as toehold-mediated strand-displacement reactions, which improve sensitivity in terms of DNA concentration but offer limited improvement in LOD [43]. LNA probe tendency to form secondary structures necessitates the use of dedicated in silico design tools, including the IDT SciTools OligoAnalyzer, for reliable primer and probe optimization [44,45].

PNA Probes

Peptide Nucleic Acid (PNA), another nucleic acid analogue, is used in a similar manner to LNA. Its backbone is composed of a chain of peptide-like N-(2-aminoethyl) glycine. PNA oligonucleotides exhibit higher melting temperatures as a result of a lack of electrical charge.

The analysis by Fouz and Apella [46] showed higher binding specificity of PNA probes for SNV detection. Assays were based on differences in melting temperature which, depending on the site and type of mismatch, reached 8–20 °C. The PNA oligonucleotides were also used in droplet digital PCR (ddPCR) as allele selective polymerization blockers, increasing sensitivity from 0.25% to 0.1% [47].

Unlike other oligonucleotide analogues, this modification prevents degradation by nucleases but is more expensive and lowers specificity. Consequently, the use of PNAs with short sequences is not advised for detecting mutations below 1% of SNV content, as they may yield false positive or false negative results [47].

OWL and MB Probes

Other SNV detection approaches include OWL (OWL1 and OWL2) and molecular beacon (MB) probes (Figure 2A–C). These methods are based on primer binding in the presence of a universal molecular beacon (UMB), leading to the formation of an “OWL” nucleic acid structure. MB probes are GC-rich, which increases melting temperature and fluorescence quenching efficiency by bringing the reporter and quencher into close proximity. Upon hybridization with a complementary target, the hairpin structure opens, generating a fluorescence signal (Figure 2A). Unlike PCR, detection is performed during an annealing step at low temperatures (5–32 °C) for up to 60 min [33,48,49,50,51].

In OWL1, primers induce opening of the MB structure, with one oligonucleotide containing a target specific base. The resulting circularized oligonucleotide configuration creates a structural lock that enhances mismatch discrimination (Figure 2B). Detection selectivity depends not only on probe–target duplex stability but also on the stability of the entire nucleic acid nanostructure, which fails to assemble in the presence of a single-nucleotide mismatch. A key advantage of OWL probes is their modular design, enabling simplified optimization and multiplexing, as target specificity is determined by primers rather than by direct probe–DNA hybridization [48,50].

OWL2 was developed to improve sensitivity and specificity by introducing an additional scaffold and extended template-binding arms, which stretch the UMB probe only in the presence of a perfectly matched target (Figure 2C) [48,50]. This approach demonstrates enhanced performance compared with OWL1, including successful SNV detection, although both OWL variants remain unsuitable for quantitative analysis [50].

2.1.2. Application of Ligases

OLA

The oligonucleotide ligation assay (OLA) was specifically developed for the determination of single-nucleotide polymorphisms and is a technique that is based on nucleic acid hybridization. The assay requires the simultaneous use of both common and variable ligation probes (Figure 3).

DNA ligase covalently joins the two oligonucleotides only if both are perfectly hybridized to their target templates, including the altered base [8,52]. Multiplexing is possible by attaching different fluorophores to probes, or by changing sequence lengths on allele-specific probes (thus changing the Tm) [8,36,52]. The lengths of flanking regions range from 15 bp to 30 bp. However, as a rule of thumb, the greater the length of the flank, the lower the selectivity. In addition, the variable probes targeting should have different melting temperatures, for higher specificity. The specificity of the assay is impacted by the SNV adjacent sequence, and consequently also by the choice of ligase itself.

Upon successful optimization, OLA can be combined with qPCR and allows for high sensitivity to be achieved qualitatively as well as quantitatively. Additionally, OLAgen software was developed to facilitate the design of ligation probes. OLA achieved a sensitivity of 1%; however, lower mutated copy content was not attempted [8,53].

LCR

Ligase chain reaction (LCR) employs four primers, two per DNA strand, which hybridize adjacently to form a ligation junction that is sealed only in the absence of a mismatch, enabling SNV discrimination. High specificity is achieved through mismatch-sensitive ligation, provided that blunt-end ligation activity is minimized to prevent target-independent ligation events [9]. LCR allows both qualitative and quantitative analysis and can be adapted for multiplexing; however, its specificity decreases when the proportion of variant alleles is low relative to the wild-type sequence [9,54]. Reported limits of detection are typically around 0.1% [54].

An advanced variant, quantitative gap-LCR (QGLCR), combines Gap-LCR with qPCR and real-time fluorescence readout (Figure 4) [9,54]. In this approach, a single-nucleotide gap between adjacent probes is filled by DNA polymerase, followed by ligation. Only the nucleotide corresponding to the target variant base is added to the reaction mixture. Using probe lengths of 15–16 nt, QGLCR can achieve limits of detection as low as 0.001%, although higher values (~0.085%) have also been reported [9,55,56]. A major limitation of this approach remains the difficulty of implementing robust multiplex designs [55,56].

LAMP

Loop-mediated isothermal amplification (LAMP) is well suited for analysis of samples with low DNA input and relies on Bst DNA polymerase, which enables rapid strand-displacement amplification without the need for thermocycling [57,58]. Due to its high efficiency and minimal equipment requirements, LAMP is particularly attractive for routine qualitative analyses in laboratories with limited technical resources [59,60,61]. However, conventional LAMP exhibits limited specificity for detecting single-nucleotide variants [57].

Specificity can be improved by introducing artificial mismatches, as implemented in artificial mismatched ligation–LAMP (AML-LAMP). This approach uses two probes: a universal phosphorylated probe and a mutation-specific probe containing an additional 3′-end mismatch to suppress nonspecific ligation (Figure 5). Ligation occurs only in the presence of the target variant, followed by standard LAMP, which can be monitored in real time using intercalating fluorescent dyes. Reported limits of detection for AML-LAMP are approximately 0.01%, while the method is generally considered suitable for qualitative rather than quantitative analysis [58].

LAMP assays may incorporate colorimetric detection using pH-sensitive dyes or metallic nanoparticles [10,62,63]. Although such approaches enable visual interpretation without instrumentation, they are prone to false-positive signals and are not compatible with multiplex analysis, limiting their applicability in routine testing [10].

RCA

The isothermal rolling circle amplification (RCA) is frequently used for specific amplification in diagnostic tests [63,64,65,66]. In addition to primers and dNTPs, RCA requires a strand displacement polymerase (e.g., Phi29) and padlock probes.

RCA can be used in two ways. The first, stand-alone RCA, uses the target DNA as a primer to amplify a previously closed padlock probe [64,67]. The second, more common, is ligation RCA (L-RCA), which uses the DNA template to close the padlock probe by ligation, forming a circular DNA structure that can be amplified. Probe closure occurs only if both ends hybridize perfectly (Figure 6).ariants of RCA include branched RCA (BRCA) and hyper-branched RCA (HRCA). BRCA uses an additional primer that binds to the RCA product and initiates synthesis of a new DNA strand, forming branched structures that increase detection sensitivity. HRCA follows the same principle but requires more primers [67,68,69,70].

Readouts include fluorescence increase in real-time or gel electrophoresis [66]. RCA products can be further analyzed quantitatively, e.g., by real-time PCR [64,71].

2.1.3. Application of CRISPR-Cas

Cas nuclease specificity is determined by guide RNA molecules. Their cleavage sites are located 2–3 nucleotides up-stream of the protospacer adjacent motif (PAM), a short (2-6nt) Cas specific sequence [72]. The effectors (Cas9, Cas12, Cas13) depend on the target sequence, as they recognize different PAMs and substrates: Cas9 and Cas12 bind and cleave dsDNA, while Cas13 targets RNA [73].

The integration of CRISPR-Cas nucleases with other methods has led to techniques such as Cas-EXPAR (exponential amplification reaction), RPA (recombinase polymerase amplification), SHERLOCK (Specific High-Sensitivity Enzymatic Reporter UnLOCKing), HOLMES (one-HOur Low-cost Multipurpose highly Efficient System) and others [73,74,75,76,77]. Cas-EXPAR combines CRISPR/Cas9 precision with the sensitivity of isothermal EXPAR, where the Cas9/sgRNA complex cuts target DNA to initiate amplification [78]. HOLMES uses Cas12a to detect DNA after amplification, employing non-specific nuclease activity for signal generation [76]. SHERLOCK uses Cas13a ribonuclease activity to degrade reporter RNA molecules following isothermal target amplification and transcription, achieving an LOD of 0.1% [73,74].

CRISPR/Cas12a- or Cas13a-based RCA can detect targets within amplified RCA products. RCA product is recognized by the Cas12a–crRNA complex through template-guide complementarity. Resulting activation of the complex triggers the non-specific trans-nuclease activity of Cas12a, cutting single-stranded DNA reporters present in the solution and generating a measurable fluorescent signal. This approach enables multiplexing and very sensitive detection, with reported LODs as low as 0.036% [79,80], which, while slightly higher than some LNA probe-based methods, remains highly effective for diagnostic applications.

2.2. Sequencing-Based Methods

SNVs can be detected directly by DNA sequencing, with all its generations commonly used for this purpose [81]. While highly precise, sequencing is labor-intensive and relatively expensive. SNVs can be identified via whole-genome sequencing (WGS) or targeted sequencing, with the latter being more commonly used due to lower costs.

Sequencing read length typically ranges from 150 to 300 bp, depending on the method. GC-rich or GC-poor sequences can introduce errors during library preparation and sequencing, especially in PCR amplification due to polymerase low fidelity. Errors can be mitigated by optimizing polymerases, thermocycling conditions, and buffers [82]. High-fidelity polymerases (e.g., Q5, Phusion, Platinum Taq) and minimizing amplification cycles help reduce errors and preserve library complexity [11].

2.2.1. Sanger

Sanger sequencing, a first-generation technology, relies on the incorporation of chain-terminating ddNTPs by DNA polymerase during amplification. Sanger sequencing is commonly used for routine analysis of a small number of targets due to its low cost and simplicity compared to next generation sequencing (NGS). It is unsuitable for quantitative analysis, requiring homogeneous samples and having a theoretical detection limit of 15–20% [83,84]. Poor performance can result from low template quality or quantity, heterogeneous samples, or sequence complexity (e.g., GC-rich or repetitive). Chromatogram analysis is important to verify sequencing quality and sample homogeneity [85].

2.2.2. Illumina

Sequencing technology developed by Illumina reads short DNA fragments. Illumina‘s sequencing by synthesis (SBS) uses fluorescently labeled nucleotides and reversible terminators, which both contribute to its high accuracy and massively parallel sequencing capabilities [86]. Library preparation involves DNA fragmentation (physical or enzymatic), end repair, and adapter ligation for PCR amplification and flow cell attachment. Illumina is a gold standard in NGS, with multiple library reagent suppliers [87]. Sequencing is robust to errors due to sample heterogeneity (99.2% specificity), and no differences in specificity towards specific substitutions (W↔S; transversions) were observed. The number of reads and depth of coverage have a key impact on the accuracy of Illumina sequencing. Low coverage increases the risk of false negatives. Conversely, too-high coverage does not guarantee a reduction in false positives—they may even increase in certain regions [88]. Fortunately, these errors can often be corrected at the data analysis stage. In summary, SNV detection based on Illumina technology achieves a LOD of 0.1% [12].

2.2.3. AVITI

AVITI is a next-generation SBS platform that generates sequence data via binding of avidites: multivalent, labeled oligonucleotide ligands. This approach differs from the likes of Illumina and Ion Torrent, where the sequencing signal is generated simultaneously to nucleotide incorporation during synthesis. First, the sequencing polonies are generated on a solid base, using rolling-circle amplification (Figure 7). Then the sequencing reaction proceeds as follows: (i) polymerase molecules bind to the 3′ ends of the multiple primer-template complexes in the polony and select the complementary avidites, allowing a fluorescence readout; (ii) avidites are removed from the polonies; (iii) template strand is extended by one nucleotide, and the cycle can be repeated [89]. Such simultaneous interaction with multiple sites increases the binding affinity 100-fold over an individual interaction. This method promises a reduction in errors and an increase in variant detection sensitivity, potentially making it an improvement over Illumina sequencing for SNV diagnostics [90,91,92,93]. Additional research would be required to verify the applicability of this platform for SNV detection [91].

2.2.4. Ion Torrent

Ion Torrent is a next-generation semiconductor chip-based sequencing technology. To generate the sequencing template, the library is clonally amplified by emulsion PCR onto beads. These beads are then placed in millions of wells connected to a semiconductor chip. Amplified DNA fragments serve as templates and are sequenced by sequentially introducing unlabeled and unmodified nucleotides in a base-by-base order. Nucleotide incorporation and the creation of a phosphodiester bond leads to a surge of protons, causing shifts in pH and electric potential, which are then detected by semiconductor chips. Since detection is direct, each nucleotide inclusion is recorded within seconds [12]. If a nucleotide is not incorporated, no voltage change is registered, which enables a simple sequence readout [94]. This technology tends to generate a lot of false positives in indel detection. Ion Torrent and Illumina have been repeatedly compared due to them generating similar error types. It has been shown that the frequency of these errors is much higher with Ion Torrent. However, upon appropriate optimization (e.g., enhancement of coverage depth), it is possible to achieve satisfactory results, with a sensitivity of 0.15% and 100% accuracy [12,95].

2.2.5. Long-Read Sequencing

Long-read sequencing (third-generation sequencing, e.g., PacBio, Oxford Nanopore Technologies), although highly useful for analyzing structural genomic variations, is not the optimal method for detecting SNVs in food and feed. Firstly, it suffers from a higher per-read error rate compared to short-read technologies [38]. Secondly, the fragmentation of template DNA in processed material renders the biggest advantage of long-read sequencing–the length of reads, ineffective. Thirdly, calling variants from mixed samples requires a high coverage of the investigated site, once again making short read sequencing more efficient for this application. Nonetheless, we predict that third-generation sequencing will have its niche for calling SNVs in homogenous samples, especially when there is little known about the possible nature of tested sites [96,97].

3. Perspectives for Adaption to GMO Control

The results reported by various independent research groups were obtained for detecting human DNA or pathogen-specific SNVs. Despite satisfactory results, these techniques may not be directly applicable in GMO enforcement laboratories for detecting SNVs in plants and food/feed products. Plant and food samples are often heterogeneous, and processed foods may contain DNA mixtures from various species, risking unwanted amplification or interference. Additionally, enzyme inhibitors such as polyphenols, tannins, starch, and lipids can reduce PCR or isothermal method efficiency, especially in seeds, further reducing SNV detection sensitivity and repeatability [98,99,100,101]. Plant DNA also has a high content of repetitive sequences and cultivar variability, potentially causing misattribution of nucleotide changes [102]. Limited information exists on these methods’ suitability for GMO analysis.

Many established methods lack sufficient characterization to meet EURL criteria, making adaptability uncertain (

Table 1. Compared the properties of SNV detection techniques.

Basic Concept	Technique	Max. LOD	LOD in GMO	Application for Quantitative Analysis	Application for Multiplex Analysis	The Biggest Drawbacks
Based on probes	LNA [40,42]	0.003%	11.2 cp (plasmid)	+	+	Tendency to form secondary structures
	PNA [47]	0.1%	N/A	N/A	N/A	Low specificity
	OWL1/2 [48,50]	N/A	N/A	N/A	+	Binding to dsDNA Difficulty
	MB [48,50]	N/A	N/A	N/A	+	Low efficiency
Based on ligation	OLA [8,53]	<1%	N/A	+	+	Low specificity
	LCR [54]	0.1%	N/A	+	+	High LOD
	QGLCR [55]	0.001%	N/A	+	+	Validation
	LAMP [58]	0.01%	N/A	−/+	+	Validation
	RCA [69]	N/A	N/A	−/+	N/A	Difficulty
CRISPR/Cas-RCA [79]		0.036%	1%	+	+	Validation

). Standard validation requirements include practicability, DNA concentration and integrity, robustness, Relative Standard Deviation of Reproducibility (RSDr), and false positive/negative rates. However, LOD and LOQ definitions differ between the fields: in medicine, LOD refers to DNA concentration in the reaction mixture [50], while in GMOs it refers to modified DNA copy number [20], reflecting differences in material availability. Thus, the adequacy of these techniques needs to be tested empirically. However, based on their known drawbacks and LODs, it is possible to delineate methods in which optimization may prove to be a Sisyphean task. Such methods, due to their high LOD, include PNA (Table 1) [47]. In this context, the use of CRISPR/Cas has not been excluded, as it can be integrated with others [79]. Quantitative analysis is not possible with published LAMP methods and with OWL1 and MB probes, which is a significant drawback in the context of official control [50,59]. Nevertheless, LAMP appears to be a suitable choice for screening, despite even the somewhat more complicated ‘initial’ procedure. It is overly optimistic to assume that all techniques could be multiplexed, which is important for high-throughput analysis [58]. In the case of RCA and OLA, the amount of information is limited, but they seem to be promising tools to increase the specificity of any novel methods [52,69].

Nonetheless, some methods have been tested for GMO and NGT analyses. Among probe-based approaches, nucleic acid analogs have been used to improve existing PCR-based assays. For example, a qPCR method with LNA primers targeting FAD2-1A and FAD2-1B variants in soybean showed promising results, with in-house validation achieving an LOD of 3.7 copies and meeting other performance criteria. Interlaboratory validation confirmed acceptable performance, though the LOD increased to 11.2 copies, likely due to plasmid-based sample variability. Overall, the results were comparable to routine methods for conventional GMOs [103].

A CRISPR/Cas12a-based method for NGT rice detection was optimized to identify SNVs at the 12-copy level, but required a specific PAM [104]. This limitation can be addressed by using modified Cas proteins, such as SpRY (PAM-less Cas9 variant), which expand target range and enable precise single-nucleotide cleavage [105,106]. Combining SpRY with recombinase polymerase amplification (RPA) allowed for the detection of single-nucleotide substitutions at the 1% level in under one hour, offering a rapid, equipment-light screening tool, though it did not meet MPR [107].

Some studies evaluated the applicability of current SNV detection methods under official GMO control. A 2020 publication by Chhalliyil et al. [108] suggested quantification of NGT products using qPCR and Sanger sequencing, but replication attempts by Weidner et al. [109] revealed non-specific amplification and failure to meet MPR criteria, with quantitative analysis hindered by the absence of established performance standards.

Targeted high-throughput sequencing enables detection of genetically edited rice lines via unique SNV markers. The workflow involves whole-genome sequencing, comparative analysis to identify unique variants, and validation with targeted PCR and NGS. This method can detect gene content below 25 template copies per sample, offering high sensitivity, accuracy, and repeatability while meeting standards analogous to MPR and EURL, thus supporting effective monitoring even at very low edited-line content [20,110].

A reproducibility study in CRISPR/Cas9-edited oilseed rape compared targeted sequencing (Illumina) and WGS (PacBio Sequel II). Long-read technology showed higher interlaboratory variability than short reads, but targeted sequencing demonstrated sufficient sensitivity to detect GMOs at the 0.1% level [111].

A key challenge in adapting SNV detection methods for routine GMO and NGT testing is the lack of standardization and harmonization across laboratories. Even highly sensitive methods may yield non-comparable results due to differences in equipment, DNA extraction reagents, or data interpretation, affecting reproducibility [20,23]. Implementing SNV detection requires standards compatible with routine procedures, including uniform definition of method parameters, standardized bioinformatic analysis, and reporting [23,112,113]. Interlaboratory studies and proficiency tests help identify weaknesses, correct protocols, and set minimum requirements [20,112]. Harmonization is especially important for high-throughput or multiplexed analyses, since not all techniques integrate easily into single analytical workflows [114].

NGT regulations vary by modification type. NGT1 plants lack mandatory SNV detection validation, while NGT2 plants follow strict criteria like conventional GMOs (sensitivity, specificity, LOD/LOQ, repeatability) [20,23,26]. Regulatory gaps complicate differentiation between natural variation and genome-edited changes, highlighting the need to strengthen analytical capabilities [20,23,26].