Systems and methods to detect rare mutations and copy number variation

What is this patent about?

’822 is related to the field of molecular diagnostics and bioinformatics, specifically focusing on the detection of rare genetic mutations and copy number variations (CNVs) within cell-free polynucleotides. The technology addresses the challenge of identifying low-abundance genetic signals, such as those derived from tumors or fetuses, which are often masked by the high noise and error rates inherent in standard next-generation sequencing workflows.

The underlying idea behind ’822 is to treat the sequencing process as a communication channel prone to noise and distortion, and to apply a systematic encoding and decoding strategy to recover the original genetic message. By attaching tags to fragmented DNA before amplification, the system creates a way to trace multiple sequence reads back to their single parent molecule, allowing the system to distinguish between true biological variants and artifacts introduced during PCR or sequencing.

The claims of ’822 focus on a method that converts a population of cell-free DNA into non-uniquely tagged parent polynucleotides using identifier sequences like barcodes. The process involves amplifying these tagged molecules, sequencing the resulting progeny, and then grouping the sequence reads into families based on shared barcodes and identical genomic start and stop positions. These families are then collapsed into a single consensus base call at specific genetic loci to determine the true frequency of mutations.

In practice, the invention functions by utilizing the diversity of the naturally occurring fragments in a sample alongside a limited set of barcodes to achieve unique identification of molecules. By collapsing reads into consensus sequences, the system effectively eliminates random errors; for instance, if a mutation appears in only one read of a family but not others, it is discarded as noise. This allows for the detection of rare variants with a sensitivity as low as 0.1%, even when the input DNA is limited to less than 100 nanograms.

This approach differs from prior methods by significantly increasing conversion efficiency—the percentage of original molecules successfully tagged and sequenced—and by using family-based filtering rather than simple quality score thresholds. While traditional sequencing might mistake a 1% error rate for a real mutation, this method uses the redundancy of amplified families to provide a high-fidelity genetic profile, enabling non-invasive monitoring of cancer progression, therapy response, and residual disease.