# US 2026/0018252 A1 · Machine learning techniques for analysis of structural variants

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

- Number: US 2026/0018252 A1
- Document ID: US20260018252A1
- Status: Published application
- Kind: A1
- Application number: 19/300,985
- Filed: 2025-08-15
- Published: 2026-01-15
- Earliest claimed: 2016-04-27
- Assignees: Arc Bio, LLC
- Applicants: Arc Bio, LLC
- Inventors: Thomas J. WATSON, JR.; Alejandro QUIROZ ZARATE
- Web page: https://patentgrep.com/patent/US20260018252A1

## CPC

- G16B 40/20: Supervised data analysis
- G06N 20/20: Ensemble learning
- G16B 15/00: ICT specially adapted for analysing two-dimensional [2D] or three-dimensional [3D] molecular structures, e.g. structural or functional relations or structure alignment
- G16B 30/10: Sequence alignment; Homology search
- G16B 40/00: ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
- G16B 30/00: ICT specially adapted for sequence analysis involving nucleotides or amino acids
- 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

## Abstract

The present disclosure provides techniques for analysis of genetic features. In particular, machine learning techniques can be used to analyze various statistical features in determining genetic features such as variants, markers, and traits, for example in a nucleotide sequence.

## Claims

### Claim at position 1

1-197. (canceled)

### Claim 198

A system comprising:
a processor in communication with a memory, the memory having stored thereon a set of instructions which, when executed by the processor, cause the processor to:
analyze, using a trained algorithm and based on a moving window, aligned reads from a nucleotide sequence for at least one statistical feature; and
determine, using the trained machine learning algorithm, a presence of a genetic feature in the nucleotide sequence based on the analyzed statistical features.

### Claim 199 (depends on 198)

The system of claim 198, wherein the trained algorithm comprises a feature obtention module, a candidate breakpoint location module, and a classification module.

### Claim 200 (depends on 198)

The system of claim 198, wherein the trained algorithm comprises a random forest algorithm.

### Claim 201 (depends on 198)

The system of claim 198, wherein the trained algorithm is trained using a moving window.

### Claim 202 (depends on 198)

The system of claim 198, wherein the moving window has a length of about 20 base pairs.

### Claim 203 (depends on 198)

The system of claim 198, wherein the moving window has a length of about 50 base pairs.

### Claim 204 (depends on 198)

The system of claim 198, wherein the moving window has a variable length.

### Claim 205 (depends on 198)

The system of claim 198, wherein the analyzing does not include portions of the aligned reads located outside the moving window.

### Claim 206 (depends on 198)

The system of claim 198, wherein the at least one statistical feature is selected from a group consisting of: number of paths or bubbles that fall within the window, number of beginnings of paths or bubbles that fall within the window, number of ends of paths or bubbles that fall within the window, number of complete sections of paths or bubbles that fall within the window, mean depth of paths or bubbles that fall within the window, significance of paths or bubbles that fall within the window, portion of a total length of each path of bubble that falls within the window, and VCF file information for each path or bubble that falls within the window.

### Claim 207 (depends on 198)

The system of claim 198, wherein the at least one statistical feature comprises at least two statistical features.

### Claim 208 (depends on 198)

The system of claim 198, wherein the at least one statistical feature comprises at least five statistical features.

### Claim 209 (depends on 198)

The system of claim 198, wherein the presence of the genetic feature is determined within a window of about 50 base pairs.

### Claim 210 (depends on 198)

The system of claim 198, wherein a start or end of the genetic feature is determined.

### Claim 211 (depends on 198)

The system of claim 198, wherein the genetic feature is a structural variant selected from a group consisting of deletions, insertions, and inversions.

### Claim 212 (depends on 198)

The system of claim 198, wherein the genetic feature is selected from a group comprising an individual subject marker, taxonomic marker, resistance marker, susceptibility marker, pathogenicity marker, or virulence marker.

### Claim 213

A non-transitory computer-readable medium having stored thereon instructions that, when executed by a processor, cause the processor to:
analyze, using trained algorithm and based on a moving window, aligned reads from the nucleotide sequence for at least one statistical feature; and
determine, using the trained machine learning algorithm, a presence of the genetic feature in the nucleotide sequence based on the analyzed statistical features.

### Claim 214 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the trained algorithm comprises a feature obtention module, a candidate breakpoint location module, and a classification module.

### Claim 215 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the trained algorithm comprises a random forest algorithm.

### Claim 216 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the trained algorithm is trained using a moving window.

### Claim 217 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the moving window has a length of about 20 base pairs.

### Claim 218 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the moving window has a length of about 50 base pairs.

### Claim 219 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the moving window has a variable length.

### Claim 220 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the analyzing does not include portions of the aligned reads located outside the moving window.

### Claim 221 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the at least one statistical feature is selected from a group consisting of: number of paths or bubbles that fall within the window, number of beginnings of paths or bubbles that fall within the window, number of ends of paths or bubbles that fall within the window, number of complete sections of paths or bubbles that fall within the window, mean depth of paths or bubbles that fall within the window, significance of paths or bubbles that fall within the window, portion of a total length of each path of bubble that falls within the window, and VCF file information for each path or bubble that falls within the window.

### Claim 222 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the at least one statistical feature comprises at least two statistical features.

### Claim 223 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the at least one statistical feature comprises at least five statistical features.

### Claim 224 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the presence of the genetic feature is determined within a window of about 50 base pairs.

### Claim 225 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein a start or end of the genetic feature is determined.

### Claim 226 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the genetic feature is a structural variant selected from a group consisting of deletions, insertions, and inversions.

### Claim 227 (depends on 213)

The non-transitory computer-readable medium of claim 213, wherein the genetic feature is selected from a group comprising an individual subject marker, taxonomic marker, resistance marker, susceptibility marker, pathogenicity marker, or virulence marker.
