# US 2022/0284984 A1 · Somatic variant calling from an unmatched biological sample > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2022/0284984 A1 - Document ID: US20220284984A1 - Status: Published application - Kind: A1 - Application number: 17/735,906 - Filed: 2022-05-03 - Published: 2022-09-08 - Earliest claimed: 2019-11-05 - Assignees: Personalis, Inc. - Applicants: Personalis, Inc. - Inventors: Patrick Jongeneel; Nicholas Phillips; Jason Harris - Web page: https://patentgrep.com/patent/US20220284984A1 ## CPC - G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection - G06N 3/045: Combinations of networks - G06N 3/09: Supervised learning - G06N 3/0985: Hyperparameter optimisation; Meta-learning; Learning-to-learn - G06N 5/01: Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound - G06N 20/20: Ensemble learning - G16B 40/20: Supervised data analysis - G16H 50/20: for computer-aided diagnosis, e.g. based on medical expert systems - G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS - G16H: HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA ## Abstract Methods for somatic variant calling from an unmatched biological samples is provided. The method can include obtaining nucleic acid sequence data corresponding to a biological sample of a subject. The method can also include aligning the nucleic acid sequence data to a reference genome. The method can also include identifying, based on the aligned nucleic acid sequence data, a set of candidate variants in said nucleic acid sequence data. The set of candidate variants may include one or more somatic variants and one or more germline variants. The method can also include, without using a nucleic acid sequencing data from a matching biological sample of the subject, processing the set of candidate variants using a trained machine-learning model to identify the somatic variants. The method can also include outputting a report that identifies the somatic variants. ## Claims ### Claim 1 A method comprising: obtaining nucleic acid sequence data of a biological sample of a subject, wherein a reference biological sample of the subject that corresponds to the biological sample is unavailable, and wherein the reference biological sample includes non-tumor cells only; aligning the nucleic acid sequence data to a reference genome; identifying, based on the aligned nucleic acid sequence data of the biological sample, a set of candidate variants in said nucleic acid sequence data, wherein said set of candidate variants includes one or more somatic variants and one or more germline variants; without using nucleic acid sequencing data of the reference biological sample of the subject, processing the set of candidate variants using a trained machine-learning model to identify the somatic variants; and outputting a report that identifies the somatic variants. ### Claim 2 (depends on 1) The method of claim 1, wherein the biological sample is a tumor sample of the subject. ### Claim 3 (depends on 1) The method of claim 1, wherein the trained machine-learning model includes a gradient boosted decision tree. ### Claim 4 (depends on 1) The method of claim 1, wherein the trained machine-learning model includes two classification models. ### Claim 5 (depends on 1) The method of claim 1, wherein the trained machine-learning model includes a filtration model. ### Claim 6 (depends on 1) The method of claim 1, wherein the trained machine-learning model includes a rescue model. ### Claim 7 (depends on 1) The method of claim 1, wherein the trained machine-learning model is trained using training data corresponding to a set of matched tumor-normal pairs. ### Claim 8 (depends on 1) The method of claim 1, wherein the trained machine-learning model is trained by tuning one or more hyperparameters via a randomized search. ### Claim 9 (depends on 1) The method of claim 1, wherein the report identifies at least one biomarker. ### Claim 10 (depends on 1) The method of claim 1, wherein the report identifies at least one prognostic marker. ### Claim 11 (depends on 1) The method of claim 1, wherein the report identifies a presence or absence of the one or more somatic variants. ### Claim 12 A system comprising: one or more data processors; and a non-transitory computer readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform one or more operations comprising: obtaining nucleic acid sequence data of a biological sample of a subject, wherein a reference biological sample of the subject that corresponds to the biological sample is unavailable, and wherein the reference biological sample includes non-tumor cells only; aligning the nucleic acid sequence data to a reference genome; identifying, based on the aligned nucleic acid sequence data of the biological sample, a set of candidate variants in said nucleic acid sequence data, wherein said set of candidate variants includes one or more somatic variants and one or more germline variants; without using nucleic acid sequencing data of the reference biological sample of the subject, processing the set of candidate variants using a trained machine-learning model to identify the somatic variants; and outputting a report that identifies the somatic variants. ### Claim 13 (depends on 12) The system of claim 12, wherein the biological sample is a tumor sample of the subject. ### Claim 14 (depends on 12) The system of claim 12, wherein the trained machine-learning model includes one or more of a gradient boosted decision tree, a filtration model, or a rescue model. ### Claim 15 (depends on 12) The system of claim 12, wherein the trained machine-learning model is trained using training data corresponding to a set of matched tumor-normal pairs. ### Claim 16 (depends on 12) The system of claim 12, wherein the trained machine-learning model is trained by tuning one or more hyperparameters via a randomized search. ### Claim 17 (depends on 12) The system of claim 12, wherein the report identifies at least one biomarker. ### Claim 18 (depends on 12) The system of claim 12, wherein the report identifies at least one prognostic marker. ### Claim 19 (depends on 12) The system of claim 12, wherein the report identifies a presence or absence of the one or more somatic variants. ### Claim 20 A computer-program product tangibly embodied in a non-transitory machine-readable storage medium, including instructions configured to cause one or more data processors to perform one or more operations comprising: obtaining nucleic acid sequence data of a biological sample of a subject, wherein a reference biological sample of the subject that corresponds to the biological sample is unavailable, and wherein the reference biological sample includes non-tumor cells only; aligning the nucleic acid sequence data to a reference genome; identifying, based on the aligned nucleic acid sequence data of the biological sample, a set of candidate variants in said nucleic acid sequence data, wherein said set of candidate variants includes one or more somatic variants and one or more germline variants; without using nucleic acid sequencing data of the reference biological sample of the subject, processing the set of candidate variants using a trained machine-learning model to identify the somatic variants; and outputting a report that identifies the somatic variants.