US 10793916

US Patent 10793916, titled "Systems and methods to detect rare mutations and copy number variation," was issued to Guardant Health Inc. The sole inventor listed is AmirAli Talasaz. The patent was filed on September 18, 2019, and granted (published) on October 6, 2020 [cite: patent/US10793916B2/en].

Abstract:
The patent describes methods and systems for detecting rare mutations and changes in DNA copy number using cell-free nucleic acids. These methods typically involve sequencing cell-free polynucleotides, attaching barcodes for identification, filtering out low-quality sequencing data, aligning the remaining sequence reads to a reference genome, quantifying the aligned reads in specific genomic regions, normalizing these counts, and then comparing them to a control sample to identify either copy number variations or rare mutations [cite: patent/US10793916B2/en].

Plain-Language Overview of Independent Claims:

• Claim 1: This claim describes a method for detecting copy number variations. It involves: 1) sequencing extracellular DNA fragments from a bodily sample, generating multiple reads for each fragment; 2) removing any reads that don't meet a set quality standard; 3) mapping the remaining reads to a reference genome; 4) counting these mapped reads within specific, predefined regions of the genome; and 5) determining copy number variation by normalizing these read counts (or counts of unique reads) within the predefined regions and comparing them to normalized counts from a control sample.
• Claim 11: This claim outlines a method for detecting rare mutations in a cell-free sample. It includes: 1) sequencing extracellular DNA fragments from a bodily sample, either by targeting specific regions (multiplex sequencing) or performing whole-genome sequencing; 2) filtering out low-quality reads; 3) mapping the reads to a reference genome; 4) identifying reads that show a variation compared to the reference at specific DNA positions; 5) calculating the proportion of variant reads to total reads at each position; 6) normalizing these proportions to identify potential rare variants; and 7) comparing these normalized variant frequencies to those from a control sample.
• Claim 21: This claim describes a method to understand the complexity (heterogeneity) of an abnormal medical condition in a patient. It involves creating a genetic profile from the patient's extracellular DNA, where this profile combines data obtained from both copy number variation and rare mutation analyses.
• Claim 22: This claim details a method that begins with original DNA fragments ("parent polynucleotides") that have unique tags attached. These tagged fragments are then amplified (copied) to produce a large number of "progeny polynucleotides." A subset of these amplified copies is then sequenced. Finally, the raw sequencing reads are processed ("collapsed") to generate more accurate "consensus sequences," with each consensus sequence representing one of the original tagged parent polynucleotides.
• Claim 33: This claim covers a method for detecting genetic variations in initial starting genetic material that has not been uniquely tagged. The method is sensitive enough to detect variations at very low frequencies (at least 5%, 1%, 0.5%, 0.1%, or 0.05%).
• Claim 36: This claim describes a method for determining copy number variation in a sample containing DNA fragments. It involves providing at least two groups of initial DNA fragments, each group mapping to a different location in a reference genome. For each group, the fragments are amplified, a subset is sequenced, and the resulting sequence reads are grouped into "families" (where each family originated from a single initial fragment). A quantitative measure of these families is then inferred for each group, and copy number variation is determined by comparing these measures between the groups.
• Claim 37: This claim describes a method for inferring the frequency of specific DNA bases (sequence calls) at a particular location in a sample. It involves amplifying initial DNA fragments from a sample, sequencing a subset of the amplified fragments, and grouping the sequence reads into families (each from a single initial fragment). For each set of initial fragments, a call frequency for one or more bases is inferred. This inference considers a confidence score for each call within a family (based on its frequency among family members) and then estimates the overall call frequency using these confidence scores.
• Claim 38: This claim details a method for communicating DNA sequence information while reducing errors. It involves taking an individual DNA molecule, encoding its sequence into a signal, transmitting this signal through a "channel" (like a sequencer), receiving a signal that may contain "noise" (incorrect calls) or "distortion" (uneven amplification), decoding the received signal to produce a clearer message with reduced errors, and then providing this corrected message to a recipient.
• Claim 42: This claim describes a computer-readable medium (e.g., hard drive, solid-state drive) containing software instructions. When a computer processor executes this code, it performs a method comprising: selecting predefined regions in a genome; accessing and counting sequence reads in those regions; normalizing these read counts across the regions; and then calculating the percentage of copy number variation within those predefined regions.
• Claim 52: This claim describes a specific chemical composition. It consists of a sample containing between 100 and 100,000 haploid human genome equivalents of cell-free DNA (cfDNA) polynucleotides, where these polynucleotides are tagged with between 2 and 1,000,000 unique identifiers.
• Claim 53: This claim describes a method for preparing a sample. It involves providing a sample that contains between 100 and 100,000 haploid human genome equivalents of cell-free DNA (cfDNA) polynucleotides, and then attaching between 2 and 1,000,000 unique identifiers (tags) to these polynucleotides.
• Claim 55: This claim describes a method involving fragmented DNA. It includes: 1) providing a sample containing many human fragmented DNA molecules; 2) determining a value 'z', which represents the average (mean, median, or mode) expected number of duplicate DNA fragments (fragments with identical start and stop positions) at any given genomic location; and 3) tagging the DNA fragments in the sample with 'n' unique identifiers, where 'n' is a number between 2 and 100,000 times 'z' (or smaller ranges like 10,000z, 1,000z, or 100*z).
• Claim 57: This claim describes a method focusing on creating accurate sequence representations. It involves taking at least one set of original DNA fragments that have been tagged. For each tagged fragment, multiple sequencing reads are generated. These multiple reads are then processed together ("collapsed") to create a single, more reliable "consensus sequence" for each unique original tagged DNA fragment.
• Claim 59: This claim describes a method for detecting genetic alterations or the extent of genetic variation in an individual. This detection is performed by sequencing cell-free nucleic acid, specifically by sequencing at least 10% of the individual's entire genome.

USPTO and CAFC Docket Search:
A search of the USPTO database confirms the details of US10793916, including its active legal status [cite: patent/US10793916B2/en]. Regarding CAFC 2026 dockets, while there is no explicit mention of an active CAFC docket for US10793916 itself in the provided information, the Google Patents legal status section notes ongoing litigation related to the patent family. This includes a PTAB case (IPR2025-01435) that was "Not Instituted - Procedural," meaning it did not proceed to a full review and thus would not be currently at the CAFC. However, multiple US District Court cases have been filed in the Delaware District Court (e.g., 1:20-cv-01580, 1:22-cv-00334, 1:24-cv-00687), which could potentially lead to appeals at the CAFC in 2026 or later, although no such specific CAFC docket is identified at this time [cite: patent/US10793916B2/en].