Paired plus-minus sequencing is an ultra-high throughput and accurate method for dual strand sequencing of DNA molecules

The paper introduces paired plus-minus sequencing (ppmSeq), a highly efficient and accurate method that partitions and amplifies both DNA strands on single beads to achieve superior duplex recovery rates and ultra-low error rates, enabling sensitive, tumor-naive circulating tumor DNA detection for clinical applications.

The Gist

Imagine you are trying to find a single, specific typo in a library containing billions of books. This is essentially what scientists do when they try to find tiny genetic mutations (called SNVs) in a person's DNA. These mutations are often the "smoking gun" for diseases like cancer, but they are incredibly rare, hiding among billions of normal letters.

The problem is that the machines used to read DNA (sequencers) aren't perfect. They make their own "typos" (errors) at a rate that is much higher than the rare mutations we are looking for. It's like trying to find a real spy in a crowd of actors who are all pretending to be spies.

The Old Way: The "Double-Check" Problem

For years, the gold standard for finding these rare mutations was a technique called Duplex Sequencing.

Think of DNA as a zipper with two sides (the Watson and Crick strands). To be sure a mutation is real, you need to see the same "typo" on both sides of the zipper. If the machine makes a mistake on just one side, you know it's an error. If it happens on both sides, it's a real mutation.

However, the old way of doing this was like trying to find a matching pair of socks in a laundry machine that only lets you grab one sock at a time. You had to throw away 90-95% of your DNA just to find the few pairs where you could read both sides. It was expensive, slow, and required massive amounts of DNA, which is a problem when you only have a tiny drop of blood (liquid biopsy) from a patient.

The New Solution: ppmSeq (Paired Plus-Minus Sequencing)

The authors of this paper, led by researchers from the New York Genome Center and Ultima Genomics, invented a new technology called ppmSeq.

The Analogy: The "Double-Decker Bus"
Imagine the old method was like taking two separate, single-person cars to the same destination, hoping they arrive together.
ppmSeq is like building a double-decker bus.

1. The Bus (The Bead): Instead of separating the two sides of the DNA zipper, ppmSeq keeps them stuck together (like a zipper that hasn't been opened).
2. The Passengers (The Strands): Both the top strand and the bottom strand of the DNA are loaded onto a single "bus" (a sequencing bead) at the same time.
3. The Ticket (The Adapter): They attach a special "ticket" to the bus that has a mismatched pattern (like a red stripe on the top and a blue stripe on the bottom).
4. The Ride: When the machine reads the bus, it sees both the red and blue stripes simultaneously.
   • If it sees both stripes mixed together, it knows, "Aha! This is a real double-stranded DNA molecule!"
   • If it sees only red or only blue, it knows, "This is just a single strand that got damaged or lost its partner." The computer can then throw that data away.

Why This is a Big Deal

1. Efficiency (The Yield)
The old methods were like fishing with a net that had huge holes; you caught very few fish (DNA pairs). ppmSeq is a net with tiny holes. They recovered 44% of the DNA pairs, compared to only 5-11% with the best previous methods. This means you get 4 to 8 times more useful data from the same amount of blood.

2. Accuracy (The Noise)
Because they can read both sides of the DNA at once, they can filter out the machine's "typos" incredibly well. They achieved an error rate so low it's like finding a needle in a haystack without accidentally picking up a piece of hay.

• The Result: They can detect a mutation that appears in only 1 out of 1 million DNA molecules.

3. Real-World Impact: Cancer Monitoring
The team tested this on patients with cancer.

• Tumor-Informed: If they know what mutations the patient's tumor has, ppmSeq can find those specific mutations in the blood at levels as low as 1 in 100,000. This is crucial for detecting if cancer is coming back after treatment (Minimal Residual Disease).
• Tumor-Naive (The Magic Trick): Even without knowing the patient's specific tumor mutations, they could detect cancer by looking for "signatures."
  • Example: Lung cancer patients who smoke have a specific pattern of DNA damage caused by tobacco. Urothelial (bladder) cancer patients often have a pattern caused by a specific enzyme (APOBEC3).
  • ppmSeq was sensitive enough to see these "tobacco signatures" or "enzyme signatures" in the blood, even without a tumor biopsy. It could even track if a patient was responding to treatment by seeing these signatures disappear or appear over time.

The Bottom Line

ppmSeq is like upgrading from a magnifying glass to a high-powered microscope that can see the entire DNA zipper at once. It makes finding rare, dangerous genetic mutations faster, cheaper, and more accurate.

This technology could revolutionize how we monitor cancer, potentially catching recurrence months before a CT scan would show a tumor, and could help us understand how aging and other diseases change our DNA over time. It turns the impossible task of finding a needle in a haystack into a routine job.

Technical Summary

1. The Problem

Distinguishing true biological single-nucleotide variants (SNVs) from sequencing and pre-analytical errors is a fundamental challenge in genomics, particularly for detecting low-frequency variants in:

• Liquid Biopsy: Detecting circulating tumor DNA (ctDNA) at very low allele fractions (e.g., minimal residual disease).
• Somatic Mosaicism: Identifying rare mutations in non-malignant tissues.

Current Next-Generation Sequencing (NGS) has error rates (~10⁻³ to 10⁻⁴) that obscure rare variants. While Duplex Sequencing (sequencing both Watson and Crick strands independently to form a consensus) is the gold standard for error correction (achieving error rates of 10⁻⁷ to 10⁻⁹), it suffers from critical limitations:

• Low Recovery Efficiency: Traditional duplex methods (e.g., NanoSeq, BotSeqS, CODEC) require massive over-sequencing to recover both strands of a single DNA molecule, often achieving only 5–11% duplex yield.
• Input Constraints: Many protocols rely on "bottlenecking" (diluting DNA) or enzymatic treatments (blunt-end restriction, dideoxy nucleotides) that are incompatible with highly fragmented cell-free DNA (cfDNA) or limit the total number of unique molecules that can be sequenced.
• Cost: The need for extreme over-sequencing makes whole-genome duplex sequencing prohibitively expensive for clinical applications.

2. Methodology: Paired Plus-Minus Sequencing (ppmSeq)

The authors introduce ppmSeq, a novel technology designed to encode both strands of a double-stranded DNA (dsDNA) molecule into a single sequencing read, thereby eliminating the need for over-sequencing to recover duplex information.

• Core Mechanism:

  • PCR-Free Library Prep: Utilizes the natural hydrogen bonds of dsDNA to keep both strands intact during library preparation.
  • Custom Adapters: Adapters contain a specific 6-base "bubble" of non-complementary sequences (e.g., AAAAAA on one strand and TTTTTT on the other).
  • Emulsion PCR (emPCR): The dsDNA is partitioned and clonally amplified on sequencing beads (Ultima Genomics platform). Because the template is not denatured prior to amplification, copies of the Watson strand and the reverse-complement of the Crick strand are generated on the same bead.
  • Single-Read Duplex: During sequencing-by-synthesis, the sequencer reads both strands simultaneously.
    • True Variants: If a mutation exists in the original dsDNA, both strands carry it, resulting in a high-quality, consistent signal.
    • Errors: If an error occurs (e.g., damage on one strand), the two strands will show conflicting signals at that position, resulting in a low base quality score that can be filtered bioinformatically.
  • Strand Identification: The custom adapter sequences allow bioinformatic tools to distinguish between:
    • Mixed Reads: Containing signals from both strands (true duplex).
    • Single-Strand Reads: Containing signals from only one strand (dropouts), which are computationally filtered out.

• Error Suppression Enhancements:

  • For genomic DNA (gDNA), the protocol incorporates dideoxy nucleotides (ddBTPs) during end-repair to block the amplification of nicked DNA, preventing the propagation of pre-sequencing artifacts.
  • A machine learning-based read-centric denoising framework (XGBoost classifier) is used to recalibrate variant quality scores based on sequence context, read features, and strand ratios.

3. Key Contributions

1. Linear Scalability: ppmSeq demonstrates that duplex recovery scales linearly with sequencing coverage and input DNA amount (tested from 1.8 ng to 98 ng), unlike bottlenecked methods.
2. Superior Yield: It achieves a 44% ± 5.5% duplex recovery rate, which is 4–8 times higher than state-of-the-art methods like NanoSeq (5%) and CODEC (11%).
3. Ultra-Low Error Rates:
   • gDNA: Achieved a residual error rate of 7.98 × 10⁻⁸ (using ddBTPs).
   • cfDNA: Achieved a residual error rate of 3.5 × 10⁻⁷, comparable to duplex sequencing but with significantly higher throughput.
4. Tumor-Informed and Tumor-Naive Detection: Validated the technology for detecting ctDNA at parts-per-million (ppm) levels without matched tumor tissue by leveraging mutational signatures.

4. Key Results

• Benchmarking Duplex Recovery:

  • ppmSeq recovered 44% of dsDNA molecules from cfDNA, compared to ~11% for CODEC and ~5% for NanoSeq.
  • The yield was positively correlated with total sequencing output (Pearson's R = 0.98).

• Error Rate Validation:

  • Using sperm gDNA (lowest mutational burden) with ddBTP blocking, ppmSeq achieved an error rate of 7.98 × 10⁻⁸.
  • In cfDNA (without ddBTPs due to fragmentation constraints), the error rate was 3.5 × 10⁻⁷, significantly lower than standard Illumina or Ultima sequencing (~10⁻⁴) and on par with duplex sequencing.

• Tumor-Informed ctDNA Detection (in silico mixing):

  • Sensitivity: Detected ctDNA at 10⁻⁴ (0.01%) tumor fraction across all cancer types, including those with low mutational burden (~4,000 SNVs).
  • High Burden Cancers: In tumors with high mutational burden (>10,000 SNVs), detection sensitivity reached 10⁻⁵ to 5 × 10⁻⁷ at 300x sequencing depth.
  • Comparison: Outperformed standard whole-genome sequencing (WGS) and matched the performance of targeted panels but with genome-wide coverage.

• Tumor-Naive (Plasma-Only) Detection:

  • Mutational Signatures: Successfully identified disease-specific signals in plasma without matched tumor tissue by analyzing trinucleotide mutation patterns.
    • Urothelial Cancer: Detected APOBEC3 signatures (SBS2/SBS13) and platinum-therapy signatures (SBS31).
    • Lung Cancer: Detected tobacco-exposure signatures (SBS4).
  • Correlation: The "de novo" APOBEC3 scores derived from plasma correlated strongly (R = 0.98) with tumor-informed measurements.
  • Clinical Monitoring: In longitudinal studies of lung cancer patients, ppmSeq detected treatment response and disease progression (via signature dynamics) that correlated with and sometimes preceded changes in CT imaging.

5. Significance and Impact

• Clinical Feasibility: ppmSeq offers a cost-efficient, scalable, and high-fidelity alternative to traditional duplex sequencing. By eliminating the need for massive over-sequencing, it makes error-corrected whole-genome sequencing (WGS) viable for clinical liquid biopsies.
• Early Detection & MRD: The ability to detect ctDNA at levels as low as 10⁻⁶ (parts-per-million) enables the detection of Minimal Residual Disease (MRD) and early-stage cancer, which is currently limited by the sensitivity of standard assays.
• Tumor-Naive Applications: The technology overcomes the major clinical barrier of requiring matched tumor tissue for monitoring. It allows for cancer detection and monitoring using plasma alone by leveraging genome-wide mutational signatures.
• Somatic Genetics: Beyond cancer, this method provides the necessary accuracy and throughput to explore somatic mosaicism in healthy tissues, potentially revolutionizing the understanding of aging and non-malignant diseases.

In summary, ppmSeq represents a paradigm shift in high-fidelity sequencing, combining the accuracy of duplex sequencing with the throughput and scalability required for routine clinical whole-genome applications.