Sequencing Highly Degraded Archival and Clinical Nucleic Acids

159 concepts110 papers2 proposals

Clinical biobanks and pathology archives house millions of tissue specimens that represent an invaluable map of human disease evolution. However, extracting high-resolution genomic and transcriptomic data from these repositories is notoriously difficult. The standard methods used to preserve these tissues—such as Formalin-Fixation and Paraffin-Embedding (FFPE)—as well as the natural delays inherent in postmortem tissue collection, severely degrade and chemically modify nucleic acids.

Historically, this degradation meant that archived clinical samples were incompatible with high-throughput sequencing platforms, which required pristine, fresh-frozen material. Overcoming this constraint requires a shift in how we capture genetic information: instead of trying to recover long, intact strands of DNA or RNA, modern molecular and computational techniques are designed to work with the fragmented pieces that survive.

The Anatomy of Preservation-Induced Degradation

To understand why standard sequencing fails on archival tissue, we must look at what happens to nucleic acids during preservation and storage.

Formalin fixation preserves tissue structure by forming chemical cross-links between proteins and nucleic acids. However, over time, this process causes severe physical fragmentation and chemical modification of RNA and DNA molecules. As detailed by Jérôme Toussaint et al., this degradation introduces significant noise into molecular assays, frequently forcing researchers to exclude a notable percentage of clinical samples from high-throughput studies.

Standard RNA-Seq:
[Poly-A Tail] ====================================> Priming & Sequencing (Requires Intact Strand)

Degraded/FFPE RNA:
[Fragment]   [Fragment]   [Fragment]   (Poly-A tail is lost or detached)

This chemical degradation creates two distinct problems for transcriptomics:

  1. Loss of Poly-A Tails: Standard RNA-sequencing techniques rely on poly-T primers that bind to the poly-A tails of mRNA molecules to initiate reverse transcription. As Craig April et al. observed, formalin fixation destroys these poly-A tails, rendering standard mRNA capture techniques useless.
  2. High Detection Thresholds: Because the genetic material is broken into tiny, irregular pieces, the minimum detectable fold-change for gene expression is inherently higher in preserved tissues than in fresh-frozen samples. This makes it exceptionally difficult to identify subtle biological variations or local gene-variant associations, a limitation noted by Alejandro Quiroz-Zárate et al..

A similar degradation process occurs in postmortem tissues. Immediate autolysis and environmental stressors like hypoxia begin to break down biomolecules the moment a tissue loses its blood supply. John T. Lonsdale et al. established that this rapid postmortem decay poses a major hurdle for large-scale genomic studies, requiring strict protocols to ensure that postmortem samples accurately reflect living physiology.

Short-Fragment Capture and Random Priming

If a sample consists entirely of short, fragmented nucleic acids, the solution is to design assays that do not require long-range integrity.

One prominent approach is the Whole-Genome DASL (cDNA-mediated Annealing, Selection, and Ligation) assay. Developed by Craig April et al., this technique bypasses the need for poly-A tails by utilizing random priming to target short RNA sequences. By focusing on small, reproducible fragments rather than the entire transcript, the assay achieves high sensitivity and reproducibility even with highly degraded FFPE inputs.

Similarly, targeted cDNA amplification protocols are used to rescue highly fragmented historical specimens. As demonstrated by Shawn Levy in an analysis of pre-cancerous lesions, these protocols use specific primer pools to amplify short, localized regions of interest. This targeted approach ensures that even if an RNA molecule is broken into dozens of pieces, the individual fragments can still be converted into sequenceable libraries, unlocking decades of archived clinical history for retrospective analysis.

Bypassing Destructive Chemical Treatments

Nucleic acid degradation is not just a byproduct of historical preservation; it can also be caused by the very laboratory treatments used to analyze them. This is a major bottleneck in epigenetics, where mapping DNA methylation has traditionally required bisulfite sequencing.

Bisulfite conversion uses harsh chemical treatments to change unmethylated cytosines into uracils, leaving methylated cytosines intact. However, this chemical reaction is highly destructive, causing severe DNA fragmentation and sample loss. For clinical applications with minute starting material—such as cell-free DNA (cfDNA) isolated from blood—this loss can ruin the diagnostic sensitivity of the assay.

To solve this, researchers have moved toward gentler, non-destructive alternatives:

  • Enzymatic Conversion: Instead of harsh chemicals, newer workflows utilize enzymatic reactions to modify DNA. Jens Füllgrabe et al. developed a method that simultaneously sequences genetic and epigenetic bases using enzymatic conversions. This approach eliminates chemical fragmentation, preserving the integrity of low-input clinical samples.
  • Inverting the Problem with Machine Learning: Rather than trying to prevent fragmentation, computational models can treat the fragmentation itself as the signal. Because the physical spacing of DNA cleavage is influenced by its epigenetic state (such as nucleosome positioning and methylation status), the degradation pattern contains hidden information. Yaping Liu et al. introduced FinaleMe, a machine learning framework that predicts DNA methylation directly from the natural fragmentation patterns of plasma cfDNA. By analyzing standard, non-destructive whole-genome sequencing data, this approach completely bypasses chemical bisulfite treatment.

Computational Denoising of Degradation Artifacts

When physical or chemical optimization during library preparation is not enough, mathematical models can be deployed downstream to clean up the data.

Degradation does not affect all transcripts equally; some RNA structures are inherently more stable than others, creating systematic biases in sequencing readouts. In studies exploring clinical biobanks, such as the sputum profiling framework discussed in the road to sputum omics 2.0, the authors demonstrated that applying quadratic programming and computational normalization to correct for RNA degradation significantly increases the sensitivity of gene expression analysis. By mathematically de-noising the data, these tools allow researchers to detect subtle, differentially expressed genes that would otherwise be obscured by the technical noise of sample decay.

Key Foundational Work

Special Paper - Introduction (2003) — contributed 2 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 113 papers
  2. Version 1 · 2026-10-04 · 85 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_ffpe_sample_limitations_3a20205e,
  title = {FFPE sample limitations},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/ffpe-sample-limitations/rdr_3a20205eb861423e8b1d315b6a561dda},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 2, frozen text):

@misc{lacuna_ffpe_sample_limitations_3a20205e_v2,
  title = {FFPE sample limitations},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/ffpe-sample-limitations/rdr_3a20205eb861423e8b1d315b6a561dda/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-19},
}
About

These directions overlap with this one through shared papers.

See something wrong? Report a mistake