# US 12,762,151 B2 · Variant pathogenicity scoring and classification and uses thereof

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 12,762,151 B2
- Document ID: US12762151B2
- Status: Granted
- Kind: B2
- Application number: 17/381,819
- Filed: 2021-07-21
- Published: 2022-01-27
- Granted: 2026-10-13
- Earliest claimed: 2020-07-23
- Assignees: Illumina, Inc.
- Applicants: Illumina, Inc.
- Inventors: Hong Gao; Kai-How Farh; Jeremy Francis McRae
- Web page: https://patentgrep.com/patent/US12762151B2

## CPC

- G06F 16/285: {Clustering or classification}
- G06N 3/02: Neural networks
- G16B 20/20: Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
- G16B 40/20: Supervised data analysis
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G16B: BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY

## Abstract

Derivation and use of pathogenicity scores for gene variants are described herein. Applications, uses, and variations of the pathogenicity scoring process include, but are not limited to, the derivation and use of thresholds to characterize a variant as pathogenic or benign, the estimation of selection effects associated with a gene variant, the estimation of genetic disease prevalence using pathogenicity scores, and the recalibration of methods used to assess pathogenicity scores.

## Claims

### Claim 1

A method for classifying a variant of interest, the method comprising:
executing, on one or more processors of a computer system, stored instructions for classifying the variant of interest, wherein the stored instructions, when executed, cause the one or more processors to perform acts comprising:
generating a pathogenicity score for a variant of interest of a gene using a neural network trained using both human sequence data and non-human primate date sequence data;
comparing the pathogenicity score for the variant of interest to a gene-specific pathogenicity threshold determined for the gene, wherein different genes have different respective pathogenicity thresholds; and
classifying the variant of interest as a pathogenic variant in response to the pathogenicity score exceeding the gene-specific pathogenicity threshold determined for the gene.

### Claim 2 (depends on 1)

The method of claim 1, wherein the pathogenicity score is based on an extent of purifying selection for the variant of interest represented in the human sequence data and non-human primate date sequence data.

### Claim 3 (depends on 1)

The method of claim 1, further comprising:
determining a correlation between mean pathogenicity scores for pathogenic variants and percentiles of pathogenicity scores;
wherein the correlation between the mean pathogenicity scores for pathogenic variants and the percentiles of pathogenicity scores is used to define the gene-specific pathogenic threshold.

### Claim 4 (depends on 1)

The method of claim 1, wherein the pathogenicity score is generated based on an amino acid sequence processed by the neural network, wherein the neural network is trained to generate the pathogenicity score from amino acid sequences.

### Claim 5 (depends on 4)

The method of claim 4, wherein a central amino acid of the amino acid sequence corresponds to the variant of interest.

### Claim 6 (depends on 1)

The method of claim 1, wherein the gene-specific pathogenic threshold is in the range defined by and including the 51st percentile to the 99th percentile.

### Claim 7 (depends on 1)

The method of claim 1, wherein the gene-specific pathogenic threshold is in the range defined by and including the 75th percentile to the 99th percentile.

### Claim 8 (depends on 1)

The method of claim 1, wherein the stored instructions, when executed, cause the one or more processors to perform further acts comprising:
comparing the pathogenicity score for the variant of interest to a gene-specific benign threshold determined for the gene, wherein different genes have different respective benign thresholds; and
in response to the pathogenicity score being less than the gene-specific benign threshold, classifying the variant of interest as benign.

### Claim 9 (depends on 8)

The method of claim 8, further comprising:
determining a correlation between mean pathogenicity scores for benign variants and percentiles of pathogenicity scores;
wherein the correlation between the mean pathogenicity scores for benign variants and the percentiles of pathogenicity scores is used to define the gene-specific benign threshold.

### Claim 10 (depends on 8)

The method of claim 8, wherein the gene-specific benign threshold is in the range defined by and including the 1st percentile to the 49th percentile.

### Claim 11 (depends on 8)

The method of claim 8, wherein the gene-specific benign threshold is in the range defined by and including the 1st percentile to the 25th percentile.

### Claim 12 (depends on 1)

The method of claim 1, further comprising:
determining the gene-specific pathogenicity threshold for the gene based on a correlation between mean pathogenicity scores for pathogenic variants of the gene and pathogenicity scores at the 51st percentile or greater of missense variants of the gene.

### Claim 13

A non-transitory computer-readable medium storing processor- executable instructions that, when executed by one or more processors, cause the one or more processors to perform steps comprising:
processing a variant of interest for a gene using a pathogenicity scoring neural network to generate a pathogenicity score for the variant of interest, wherein the pathogenicity scoring neural network is trained using both human sequence data and non-human primate date sequence data;
comparing the pathogenicity score for the variant of interest to a pathogenicity threshold specific to the gene and a benign threshold specific to the gene, wherein different genes have different respective pathogenicity thresholds and benign thresholds; and
classifying the variant of interest as pathogenic, benign, or neither pathogenic or benign based on the comparison of the pathogenicity score to the pathogenicity threshold specific to the gene and the comparison of the pathogenicity score to the benign threshold specific to the gene.

### Claim 14 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein the variant is classified as pathogenic in response to the pathogenicity score exceeding the pathogenicity threshold specific to the gene.

### Claim 15 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein the variant is classified as benign in response to the pathogenicity score being below the benign threshold specific to the gene.

### Claim 16 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein the variant is classified as neither pathogenic nor benign in response to the pathogenicity score being between the benign threshold specific to the gene and the pathogenic threshold specific to the gene.

### Claim 17 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein comparing the pathogenicity score for the variant of interest to the pathogenicity threshold specific to the gene and the benign threshold specific to the gene comprises comparing the pathogenicity score for the variant of interest to the pathogenicity threshold prior to the benign threshold, wherein comparison to the benign threshold is contingent upon the outcome of the comparison to the first pathogenicity threshold.

### Claim 18 (depends on 13)

The non-transitory computer-readable medium of claim 13, wherein comparing the pathogenicity score for the variant of interest to the pathogenicity threshold specific to the gene and the benign threshold specific to the gene comprises comparing the pathogenicity score for the variant of interest to the pathogenicity threshold and to the benign threshold in parallel.

### Claim 19

A variant classification system, comprising:
one or more memory structures encoding processor-executable instructions;
one or more processors in communication with the one or more memory structures, wherein the one or more processors execute the instructions to perform operations comprising:
generating a pathogenicity score for a variant of interest of a gene using a neural network trained using both human sequence data and non-human primate date sequence data;
comparing the pathogenicity score for the variant of interest to a pathogenicity threshold specific to the gene and a benign threshold specific to the gene, wherein different genes have different respective pathogenicity thresholds and benign thresholds; and
classifying the variant of interest as pathogenic, benign, or neither pathogenic or benign based on the comparison of the pathogenicity score to the pathogenicity threshold specific to the gene and the comparison of the pathogenicity score to the benign threshold specific to the gene.

### Claim 20 (depends on 19)

The variant classification system of claim 19, wherein the variant of interest is classified as pathogenic in response to the pathogenicity score exceeding the gene specific pathogenicity threshold specific to the gene.

### Claim 21 (depends on 19)

The variant classification system of claim 19, wherein the variant of interest is classified as benign in response to the pathogenicity score being below the benign threshold specific to the gene.