Duplex Sequencing

The core innovation of Duplex Sequencing is the generation of a Duplex Consensus Sequence (DCS) . After ligating complementary Unique Molecular Identifiers (UMIs) to both ends of a double-stranded molecule, each strand is sequenced independently.

• The two resulting reads are compared base-by-base.
• A mutation is only called if it is present on both strands.
• This eliminates asymmetric errors caused by oxidative damage or polymerase mistakes that affect only one strand.

Duplex Sequencing reduces sequencing error rates from the typical 1% (1 in 100) down to approximately 1 in 10^7 or lower.

• Standard NGS relies on redundant read depth to overcome noise.
• Duplex Sequencing achieves single-molecule sensitivity.
• It can reliably detect a single true mutation among tens of thousands of wild-type molecules, making it ideal for Minimal Residual Disease (MRD) detection.

Before forming the final duplex consensus, the process often involves an intermediate step: Single-Strand Consensus Sequences (SSCS) .

• Reads sharing the same UMI are grouped into families.
• A consensus is built for each original strand to correct for PCR amplification errors and early sequencing cycles.
• The final comparison of the two SSCSs (one from the forward strand, one from the reverse) produces the high-fidelity DCS.

A critical advantage is the ability to distinguish true somatic mutations from base damage artifacts like cytosine deamination or 8-oxoguanine.

• Chemical damage typically affects only one strand of the DNA helix.
• A polymerase encountering a damaged base may incorporate a wrong nucleotide, creating a false-positive mutation call in standard sequencing.
• Because the complementary strand remains undamaged, Duplex Sequencing correctly identifies these events as artifacts and rescues the true base call.

By counting unique UMI families rather than raw read counts, Duplex Sequencing provides absolute quantification of input molecules.

• This eliminates the stochastic noise of PCR amplification.
• The variant allele frequency (VAF) directly reflects the true proportion of mutant molecules in the original sample.
• This is essential for tracking dynamic changes in circulating tumor DNA (ctDNA) levels during therapy.

The primary technical trade-off is the reduced conversion efficiency of input molecules.

• The ligation of dual UMIs and the requirement for both strands to be successfully sequenced means a significant fraction of input DNA is lost.
• Typical conversion rates range from 10% to 50%.
• This necessitates higher input amounts or deeper sequencing to achieve the desired limit of detection, a key consideration for low-biomass liquid biopsy samples.

A random nucleotide barcode ligated to individual DNA molecules before amplification. In Duplex Sequencing, complementary UMIs are attached to both strands, enabling computational grouping of reads derived from the same original duplex. This allows the construction of a single-strand consensus sequence (SSCS) from each strand independently, which is the prerequisite for the final duplex comparison.

The intermediate error-reduction product formed by collapsing reads sharing the same UMI and strand orientation. SSCS generation eliminates random polymerase errors introduced during early PCR cycles. However, it cannot resolve DNA base damage (e.g., cytosine deamination to uracil) that occurred in vivo or during sample preparation, as the lesion is present on the original template strand.

The final, ultra-high-fidelity product created by comparing the SSCS from the forward strand against the SSCS from the reverse strand. A true mutation must be present in both complementary strands at the same position. This step eliminates asymmetric errors, such as 8-oxoguanine lesions or spontaneous deamination events, which manifest as mutations on only one strand. DCS accuracy can exceed 1 error per 10^7 bases.

A synthetic nucleotide sequence incorporated into library adapters to uniquely tag individual starting molecules. In Duplex Sequencing, barcodes are designed with degenerate or semi-degenerate sequences to generate a sufficiently large combinatorial space, ensuring that no two original molecules share the same tag. This prevents barcode collision, which would cause distinct molecules to be erroneously collapsed into a single consensus.

A bioinformatic strategy leveraging redundant sequencing and molecular tags to build consensus sequences. Duplex Sequencing is the most stringent form, requiring bidirectional concordance. This contrasts with simpler read-family collapse methods that only suppress random errors but remain vulnerable to systematic artifacts like FFPE-induced cytosine deamination, which Duplex Sequencing explicitly resolves.

The lowest concentration of analyte reliably distinguishable from background noise. Duplex Sequencing dramatically lowers the LoD for somatic variants by suppressing the background error rate of conventional NGS. While standard sequencing struggles below 1% Variant Allele Frequency (VAF), Duplex Sequencing can reliably detect mutations at <0.1% VAF, enabling sensitive minimal residual disease (MRD) monitoring from liquid biopsies.