Unlike conventional PCR-based assays, which can only find pathogens for which specific primers have been designed, metagenomic next-generation sequencing, or mNGS, offers an untargeted approach. It sequences all genetic material in a clinical sample, then uses computational methods to distinguish viral reads from the vast background of human and bacterial DNA and RNA. That agnostic strategy is what makes the technology so powerful for outbreak investigation, for discovering novel viruses, and for identifying unexpected pathogens in severely ill patients. But it also comes with a fundamental sensitivity problem: in respiratory samples, viral nucleic acids are typically present in vanishingly small quantities relative to host material, and the review makes clear that what happens to a specimen before sequencing largely determines whether the virus is found at all.

The review found that sample enrichment—the process of concentrating viral particles or viral nucleic acids while discarding host and bacterial material—was the single most variable step across the 66 studies. The most frequently used technique was centrifugation, employed in roughly half of the studies, which separates viruses from larger cells and debris based on size and density differences. Enzymatic treatment was the second most common approach, used in about 45% of studies; this typically involves treating samples with DNases and RNases to digest exposed host and bacterial nucleic acids while leaving genetic material protected inside intact viral capsids untouched. Other enrichment strategies reported in the literature include filtration, host cell depletion using chemicals such as saponin, and membrane-based concentration. The choice of method is far from trivial: each one imposes a bias on which viruses can be recovered. DNase treatment, for example, effectively eliminates naked viral nucleic acids and can therefore miss viruses that do not have protected genomes at the stage of collection.

Following enrichment, most workflows rely on some form of amplification to generate enough material for sequencing, because the amount of viral RNA or DNA in a typical respiratory specimen is far below the input requirements of modern sequencers. The review identified sequence-independent single-primer amplification, known as SISPA, and SMART, or switching mechanism at the 5′ end of RNA template, as the two most commonly used amplification methods across the analyzed studies. SISPA works by attaching a known primer sequence to random fragments of nucleic acid and then amplifying everything with a single primer, while SMART uses a template-switching reverse transcriptase to add universal priming sites to the ends of RNA molecules, enabling full-length cDNA amplification. According to the review, these approaches produced better genome coverage for most viruses than the alternatives, meaning that researchers and clinicians were more likely to recover near-complete viral genomes rather than just fragments—a distinction that matters enormously for downstream applications such as phylogenetic analysis, drug-resistance detection and vaccine matching.

Other amplification strategies assessed in the review include whole transcriptome amplification, multiple displacement amplification, which preferentially amplifies circular DNA genomes, and VIDISCA, a restriction-enzyme-based technique originally developed for virus discovery. The authors note that each method carries a distinct bias profile: some favor small circular DNA viruses, others preferentially recover RNA viruses with high sequence identity to the priming oligos used. Because the biases introduced at the amplification step directly shape which parts of the viral genome are sequenced, the review argues that amplification choice should be treated as a primary design decision rather than a laboratory afterthought.

Perhaps the most clinically significant finding is how mNGS performs when compared head-to-head with routine molecular diagnostics. Across the studies that reported comparison data, agreement with conventional tests varied widely, with sensitivity ranging from 62.5% to 95%. The variability reflects differences in wet-laboratory protocol, sequencing depth, bioinformatic pipelines and, crucially, the viral load of the specimens tested. At the low copy numbers typical of late-stage or treated infections, enrichment and amplification losses compound, and mNGS can miss pathogens that a highly optimized quantitative PCR assay would catch. Conversely, at high viral loads, mNGS frequently matches or exceeds the performance of targeted panels—and it can do so without knowing in advance which virus to look for.

The review also documented patterns in which viruses were detected in which types of samples, offering a portrait of respiratory virology through a metagenomic lens. Rhinovirus was disproportionately detected in samples from patients with lower respiratory tract infections, appearing in 68% of the relevant detections, an observation that aligns with growing recognition of rhinovirus as a significant pathogen in exacerbations of asthma and chronic obstructive pulmonary disease rather than merely a cause of the common cold. Respiratory syncytial virus, by contrast, was found predominantly in upper respiratory tract specimens, accounting for 60% of its detections. Cytomegalovirus and human parainfluenza viruses also featured prominently among the detected pathogens. The authors suggest that such distribution data, accumulated across large metagenomic datasets, may eventually refine our understanding of which parts of the respiratory tract each virus colonizes and injures.

Despite these capabilities, the review is candid about why mNGS has not displaced PCR panels in hospital laboratories. Turnaround time remains a major barrier: while a multiplex PCR panel can return results within hours, a metagenomic workflow involving extraction, enrichment, amplification, library preparation, sequencing and bioinformatic analysis can take one to three days, often too slow to change clinical management of an acutely ill patient. Cost is another obstacle, both in reagents and in the substantial computational infrastructure and expertise required to analyze sequencing data. Contamination control presents a further challenge, since the extraordinary sensitivity of the method means that low-level environmental or reagent contamination can produce spurious findings, and background viral sequences introduced during sample handling or laboratory processing can be mistaken for genuine infections.

The review also highlights quality-control gaps in the published literature. Studies vary enormously in how they report sequencing depth, extraction efficiency, limit of detection and confirmation of findings, making it difficult to compare performance across laboratories. The authors argue that standardized reporting frameworks, benchmark reference materials and validated bioinformatic pipelines will be essential if mNGS is ever to be accredited as a diagnostic rather than used primarily as a research tool. Some of these gaps are already being addressed through external quality-assessment schemes and the increasing availability of reference standards spiked with known quantities of synthetic viral material.

Looking forward, the review identifies several technological developments that could accelerate the integration of mNGS into routine diagnostics. Third-generation nanopore sequencing platforms, which can generate long reads in real time and be deployed in compact benchtop formats, are already shortening turnaround times in some clinical settings. CRISPR-based detection methods coupled to sequencing outputs, improvements in uracil-DNA glycosylase and dUTP-based contamination prevention systems, and automated library-preparation robotics all feature in the review’s discussion of emerging directions. The authors also point to hybrid approaches in which metagenomic sequencing serves as a broad surveillance net, flagging unusual pathogens that are then confirmed and quantified by targeted assays—effectively combining the breadth of mNGS with the speed and sensitivity of qPCR.

For the researchers behind the review, based at the Manipal Institute of Virology in India together with collaborators at Nitte University and the UK Health Security Agency, the central message is one of cautious optimism. Metagenomic next-generation sequencing has already transformed how new respiratory viruses are discovered and how outbreaks are characterized, as demonstrated vividly during the SARS-CoV-2 pandemic when genome sequencing became a global public-health utility. What remains is the harder translational task: converting a powerful research technology into a dependable, affordable and standardized clinical tool. The review’s synthesis of 66 studies suggests that the path forward runs through careful optimization of the least glamorous steps—how samples are centrifuged, digested and amplified—as much as through advances in the sequencers themselves. As the authors conclude, overcoming these technical limitations through optimized sequencing protocols and greater cost-effectiveness will be pivotal in determining whether mNGS fulfills its promise as a routine frontline diagnostic for respiratory viral disease.