Methods and systems for detecting genetic variants

What is this patent about?

’366 is related to the field of molecular biology and medical diagnostics, specifically the detection and quantification of genetic variants such as copy number variations (CNV) and single nucleotide variants (SNV). The technology is particularly useful for analyzing cell-free DNA (cfDNA) in bodily fluids to monitor diseases like cancer, where rare genetic alterations may be present at extremely low concentrations.

The underlying idea behind ’366 is that standard sequencing often fails to capture all molecules in a sample, leading to inaccurate quantification of genetic material. By using duplex tags to differently label the Watson and Crick strands of double-stranded DNA, the system can track which original molecules were fully recovered as pairs, which were partially recovered as singlets, and—crucially—statistically infer the number of molecules that were completely missed by the sequencer.

The claims of ’366 focus on a method for detecting double-stranded DNA by tagging molecules with a significant excess of molecular barcodes to ensure high conversion efficiency. The process involves grouping raw reads into families based on barcode and fragment end sequences, collapsing these into consensus reads, and then using the observed counts of paired and unpaired strands to calculate a quantitative measure of the total original molecules, including those for which neither strand was detected.

In practice, the invention utilizes a library of adapters, such as Y-shaped or bubble-shaped tags, to ensure that complementary strands are uniquely identifiable after amplification. By comparing the sequence of one strand with its complement, the system can filter out artifacts introduced during PCR or sequencing. If a variant appears in both strands of a duplex pair, it is called with high confidence, whereas variants appearing in only one strand are flagged as potential errors.

This approach differs from prior methods by moving beyond simple redundancy reduction to address the problem of unseen molecules. While traditional bioinformatics can reduce noise for molecules that are successfully sequenced, ’366 uses the ratio of paired to unpaired reads to correct for sampling bias across different genomic loci. This statistical correction allows for much higher sensitivity and specificity, enabling the detection of rare mutations at concentrations below 1% with greater than 99.9% specificity.