In the study of genomics and infectious disease, we often look where the light is brightest. If we use DNA sequencing, we see the blueprint of the cell and DNA-based pathogens. If we use RNA sequencing (transcriptomics), we see which parts of that blueprint are currently being "read."
However, a recurring theme across recent research is that this binary choice creates a massive detection gap. Whether we are mapping the viruses in the deep ocean or identifying mutations in a patient with Alzheimer’s, relying on a single modality—either DNA or RNA—leaves significant biological "dark matter" unobserved.
The Metagenomic Blind Spot for RNA Viruses
The most immediate consequence of this gap is the systematic undercounting of RNA viruses. In clinical settings, Asha Rani and colleagues demonstrated in A diverse virome in kidney transplant patients contains multiple viral subtypes with distinct polymorphisms that standard shotgun sequencing workflows targeting DNA completely overlook the RNA virome. This isn't just a taxonomic oversight; it has life-or-death implications.
In the ICU, Mingqiang Wang found that DNA-based metagenomics fails to detect RNA viruses that drive severe pneumonia, leading to incomplete microbial profiles (Identifying subgroup of severe community-acquired pneumonia based on clinical metagenomics, a multicenter retrospective cohort study). Similarly, Haopu Yang showed that while microbial DNA translocates from lungs to blood in 40% of pneumonia cases, DNA-optimized platforms cannot capture the dynamics of RNA viruses, which are equally critical drivers of systemic inflammation (Microbial lung-to-blood translocation associates with systemic inflammation in severe pneumonia: evidence from paired plasma and lower respiratory tract metagenomics).
This bias extends to global ecology. By mining metatranscriptomes rather than just DNA, Uri Neri and Yuri I. Wolf achieved a five-fold expansion of the known global RNA virome, revealing that RNA bacteriophages are far more dominant than previously thought (A five-fold expansion of the global RNA virome reveals multiple new clades of RNA bacteriophages). Without looking at RNA, researchers like Min Wang argue that our understanding of extreme environments, like the Challenger Deep, remains fundamentally skewed (Unveiling the Ocean Viromes of Challenger Deep, the Deepest Place in the World’s Oceans).
Why RNA-Seq Misses Silent Integrations
While DNA sequencing misses RNA viruses, the reverse is also true: RNA sequencing is blind to "silent" DNA. This is a major hurdle in cancer research. Viruses like HPV and HBV drive cancer by integrating their DNA into the host genome.
Irene Kraus Christiansen found that HPV preferentially integrates into transcriptionally active regions (Transcriptionally Active Regions Are the Preferred Targets for Chromosomal HPV Integration in Cervical Carcinogenesis). If a researcher uses only RNA-seq to find these integration sites, they will only see the "loud" ones—those that produce fusion transcripts. As Seungyeul Yoo noted in the context of liver cancer, RNA-seq cannot detect latent or silent integrations that aren't being expressed, creating a biased map of the viral-host landscape (A pilot systematic genomic comparison of recurrence risks of hepatitis B virus-associated hepatocellular carcinoma with low- and high-degree liver fibrosis).
To solve this, Che-Yu Lee developed Venus, a pipeline that attempts to bridge this gap by using a massive viral index to catch even rare chimeric reads in single-cell data, though the authors acknowledge that DNA-based sequencing is still required to capture the full landscape of non-coding insertions (Venus: An efficient virus infection detection and fusion site discovery method using single-cell and bulk RNA-seq data).
The Expression Bias in Somatic Mutations
The gap also complicates our understanding of non-viral diseases. In Alzheimer’s research, Maria Kousi and Hansruedi Mathys discovered that specific brain cells accumulate non-inherited "somatic" mutations (Single-cell mosaicism analysis reveals cell-type-specific somatic mutational burden in Alzheimer’s Dementia). However, detecting these mutations via transcriptomics is inherently limited to genes that are "turned on." Any mutation in a regulatory region or a silent gene remains invisible.
New computational frameworks like RESA, developed by Tianyun Zhang, attempt to mitigate this by using machine learning to identify mutations directly from single-cell RNA-seq data (De novo identification of expressed cancer somatic mutations from single-cell RNA sequencing data). While RESA increases precision, the fundamental constraint remains: you cannot sequence what is not expressed.
Closing the Gap
The research suggests that the future of the field lies in multi-compartment and modified sequencing. For example, Julie Frouard developed HIV-seq to overcome a specific technical gap: standard scRNA-seq misses viral RNA that lacks a poly-adenylated tail. By modifying the primers, they increased detection of rare HIV-active cells by 44% (HIV-SEQ Reveals Global Host Gene Expression Differences Between HIV-Transcribing Cells from Viremic and Suppressed People with HIV).
As Holger Sültmann and the PCAWG Consortium demonstrated in their massive survey of 2,600 cancer genomes, the most complete picture emerges only when we account for the systematic biases in our extraction and sequencing protocols (The landscape of viral associations in human cancers). Moving forward, the goal is to stop choosing between the "blueprint" (DNA) and the "activity" (RNA) and instead develop tools that can see both simultaneously.