Library Prep Strategy Found to be Critical Determinant of RNA Virus Sensitivity
When scientists attempt to identify viruses in medical samples using DNA sequencing, the way they prepare the sample matters immensely. Current clinical metagenomics—the process of sequencing all nucleic acids in a specimen to identify pathogens without needing to grow them in a lab—often struggles with RNA viruses. This difficulty arises because RNA must first be converted into complementary DNA (cDNA) before it can be sequenced. While many modern protocols claim to be "pathogen agnostic" (capable of detecting any type of organism), recent research suggests they may inadvertently blind themselves to RNA viruses by choosing the wrong molecular architecture during library preparation.
The Problem
The fundamental obstacle in clinical metagenomics is the overwhelming presence of host material. In a typical clinical specimen, the ratio of human nucleic acids to pathogen nucleic acids is staggeringly high. To find a needle in a haystack, researchers use various "host depletion" strategies. These include treating samples with DNase (an enzyme that degrades DNA) to remove extracellular DNA or using filtration to remove human cells.
Even after depletion, a secondary problem persists: the sheer abundance of host RNA. Human samples are dominated by ribosomal RNA (rRNA) and transfer RNA (tRNA). These molecules are significantly shorter than the long, intact genomic RNA of most respiratory viruses. Standard library preparation protocols often utilize "tagmentation." This is a method where transposase enzymes simultaneously fragment DNA and attach sequencing adapters. While tagmentation is fast and cost-effective, the authors of this paper demonstrate that it is catastrophic for RNA virus sensitivity. By fragmenting the cDNA into a uniform distribution of short pieces, tagmentation destroys the natural size differential between long viral genomes and short host fragments. This effectively erases the viral signal in a sea of human debris.
How It Works
The researchers propose that the secret to high-sensitivity viral detection lies in exploiting a phenomenon called the PCR suppression (PS-) effect. This mechanism relies on the specific geometry of single-primer amplification. This technique is known as Sequence-Independent, Single-Primer Amplification (SISPA).
The mechanism functions through three interconnected stages:
- Adapter Incorporation: Instead of fragmenting the DNA later, adapters are incorporated directly during cDNA synthesis. This creates molecules with inverted terminal repeats (ITRs). These are identical adapter sequences at both ends of the strand.
- Intramolecular Folding: During the denaturation (separation of DNA strands) and annealing (re-joining of strands) steps of PCR, these single-stranded molecules can fold back on themselves. Because the ends are identical, they form stable "panhandle" or hairpin structures.
- Size-Dependent Selection: This is the core of the enrichment. Shorter fragments, such as host tRNA or degraded rRNA, form these panhandles very easily. This physically blocks the PCR primers from binding. As a result, these short fragments fail to undergo exponential amplification. In contrast, much longer viral molecules are statistically less likely to form stable, complete panhandles. Consequently, the longer viral templates remain accessible to primers. They are preferentially amplified, effectively "filtering" the library by length.
As shown in, the study systematically compared this SISPA-based approach against tagmentation-based methods.
The goal was to see if disrupting this length-dependent mechanism would eliminate viral recovery.
Numbers
The impact of these architectural choices is stark. In a controlled environment using the Zeptometrix Respiratory 2.1 panel (containing 16 RNA and 3 DNA viruses), the authors found that SISPA-based methods were remarkably successful. Specifically, the SMART-9N/Tag/NEB condition recovered 16 out of 16 viral genomes. The SMART-9N/Full/NEB condition recovered 15 of 16 [Table 1].
In contrast, the performance of tagmentation-based amplification was a total failure in the control setting. Every single condition that relied on the RLB primer—which targets adapters introduced via tagmentation—recovered zero viral genomes. Even when looking at the total sequencing yield, the disparity was massive. SISPA-based conditions yielded hundreds of Megabases (Mbp) of data. Meanwhile, RLB-amplified tagmentation conditions produced negligible yields, often just a few Mbp [Table 1].
The researchers also quantified the "breaking point" of this enrichment by spiking viruses into clinical samples with varying host backgrounds. At low background, SMART-9N/Full/NEB maintained a 17% viral yield and recovered 15 genomes. This demonstrates the method's ability to pull a signal out of a noisy sample. However, as the background increased to "medium," the viral yield collapsed to 1%. Only 5 genomes were recovered. At "high" background, the suppression effect reached its limit. No condition, regardless of the method, could recover any viral genomes.
What's Missing
While this study provides a mechanistic roadmap for optimizing viral detection, it is not a universal solution.
First, the PCR suppression effect is inherently biased against size. The authors admit that viruses with small genomes, such as parvoviruses, may be caught in the "suppression trap" alongside host fragments. Similarly, multi-segmented viruses like Influenza might not benefit from length-based enrichment. This is because their individual segments may be too short to escape suppression.
Second, there is a trade-off regarding bioinformatic complexity. Because single-primer PCR relies on strands folding and re-annealing, it naturally produces a higher rate of chimeric reads. These are sequences where two different DNA templates are fused together. This necessitates the use of specialized, chimera-aware bioinformatic pipelines to avoid errors in genome assembly.
Finally, the study uses a standardized mock community. While the Zeptometrix panel is a rigorous benchmark, it consists of inactivated viruses in a controlled buffer. This lacks the biochemical complexity and enzymatic degradation found in real human clinical specimens.
Should You Prototype This
If you are developing a clinical metagenomics pipeline where RNA virus detection is a primary requirement, the answer is yes. However, you must move away from standard tagmentation-based workflows.
Do not use direct tagmentation of cDNA for RNA virus discovery. Such methods are mathematically destined to fail by destroying the length-differential required for enrichment. Instead, implement a SISPA-based approach. The authors suggest prioritizing the SMART-9N protocol for its efficiency and ability to generate longer cDNA. If you adopt these single-primer methods, ensure your software stack includes robust tools for handling chimeric reads. For those looking to implement these findings, the authors have made their code notebooks and processed data available at https://github.com/bede/2026-protocol-optimisation and https://zenodo.org/records/20185851.