Systems and methods to detect rare mutations and copy number variation

What is this patent about?

’663 is related to the field of molecular diagnostics and bioinformatics, specifically focusing on the high-sensitivity detection of genetic aberrations in cell-free polynucleotides. The technology addresses the challenge of identifying rare mutations and copy number variations (CNVs) that are often masked by the inherent noise and distortion produced during PCR amplification and high-throughput sequencing. By analyzing DNA fragments shed into bodily fluids, the system provides a non-invasive means to monitor disease states such as cancer or fetal abnormalities.

The underlying idea behind ’663 is the use of a digital sequencing framework that treats polynucleotide sequencing as a communication theory problem, where the original molecule is a message and sequencing artifacts are noise. The key inventive insight lies in a non-unique tagging strategy combined with physical fragment characteristics—specifically start and stop positions—to track progeny molecules back to their original parent strands. This allows the system to collapse multiple sequencing reads into high-fidelity consensus sequences, effectively filtering out stochastic errors and amplification biases that would otherwise obscure low-frequency somatic variants.

The claims of ’663 focus on a method for detecting somatic genetic variants by ligating adapters containing molecular barcodes to both ends of cell-free DNA (cfDNA) molecules. Crucially, the method employs a specific tagging ratio where the number of unique barcodes is at least two but fewer than the total number of cfDNA fragments mapping to a specific genomic position. The independent claims further specify the use of these barcodes in conjunction with mapping coordinates (start and stop positions) to group sequencing reads into families, enabling the detection of single nucleotide variants, indels, gene fusions, and copy number variations.

In practice, the invention works by extracting cfDNA from a subject's blood or other bodily fluid and converting these fragments into a tagged library with high efficiency. After amplification and optional enrichment for cancer-related target regions, the resulting progeny molecules are sequenced. The bioinformatics pipeline then organizes these reads into familial groups based on their barcodes and genomic alignment. By comparing members within each family, the system can distinguish a true biological mutation—which would appear across the majority of family members—from a sequencing error that appears only sporadically.

This approach differs from prior methods by moving beyond simple quality filtering of individual reads to a structured signal decoding process at the family level. While traditional sequencing often requires large amounts of input DNA to overcome low conversion rates, this invention utilizes optimized ligation and digital collapsing to achieve high sensitivity even with limited samples. By normalizing read counts and utilizing consensus sequences, the system provides a more accurate quantitative measure of genetic material, allowing for the detection of rare variants at frequencies as low as 0.1% or less.