K-mer Variant Calling With De Bruijn Graphs for Faster Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for identifying genomic or transcriptomic sequence variants require extensive processing power and time, making them unsuitable for delivering personalized medical treatment on a meaningful scale.

Innovation Solution

A computer-implemented method for variant calling using a de Bruijn graph format to break sequence data into fragments, match them against a database, and identify variants with the highest k-mer count, utilizing a database that includes reference and variant sequence information.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional methods are used to identify genomic variants, then measurement precision is improved, but processing time and computational resources increase significantly

Engineering Contradiction:
Improvevariant identification accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent segments the genomic sequence into k-mers (short substrings of length k) and represents them as nodes in a de Bruijn graph. This segmentation allows the complex task of variant identification to be broken down into manageable units, enabling parallel processing and significantly reducing computational time while maintaining accuracy through the graph-based assembly of these segments.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The de Bruijn graph serves as an intermediary data structure between the raw sequence data and the variant identification process. By transforming sequences into graph representations where nodes are k-mers and edges represent overlaps, the system creates an intermediate representation that facilitates efficient comparison and variant detection without requiring direct processing of entire genomic sequences.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Measurement precision

If traditional methods are used to identify genomic variants, then measurement precision is improved, but processing power requirements increase

Engineering Contradiction:
Improvevariant identification accuracyVSAvoidprocessing power
Core Design Contradiction:
Measurement precisionVSPower

Solution Approach 1:

By dividing the genomic sequence into k-mers and representing them as graph nodes, the patent reduces the computational complexity per operation. The graph structure allows for efficient traversal and comparison using algorithms that scale better with data size, significantly reducing the processing power required compared to traditional alignment-based methods that must handle entire sequences.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the fundamental parameter of data representation from raw sequence strings to graph-structured k-mer assemblies. This parameter change enables the system to maintain high variant identification accuracy while reducing computational resource requirements through more efficient data structures and algorithms that operate on the graph representation rather than raw sequences.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If extensive processing power and time are used for variant calling, then variant identification accuracy is improved, but scalability for personalized medical treatment is reduced

Engineering Contradiction:
Improvevariant identification accuracyVSAvoidscalability
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The segmentation of sequences into k-mers and representation as graph nodes enables the system to process multiple samples in parallel. The graph-based approach allows for efficient handling of multiple genomes simultaneously, as the same graph structure can be used across different samples, significantly improving scalability for population-level personalized medicine applications.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The de Bruijn graph representation serves as a universal data structure that can handle various types of genomic data and variant types within a single framework. This multi-functional approach allows the system to scale efficiently by using the same computational infrastructure for different applications, from single-sample analysis to population-wide screening, without requiring proportional increases in processing power.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentEP3935638B1System and method for variant calling
Publication Date: 2025.11.05 NANTOMICS LLC
  • EP3935638B1 patent drawingFigure 1
  • EP3935638B1 patent drawingFigure 2
  • EP3935638B1 patent drawingFigure 3

AI summary

A locus tester or locust database has stored therein DNA or RNA sequence information for one or more loci of interest. The sequence information may include a list of k-mers in a given DNA or RNA sequence, an identification of whether each k-mer in the list of k-mers appears in a reference sequence or in a variation of the reference sequence, and a count of how many times each k-mer in the list of k-mers has been identified in sequence information for the locus of interest in question. Sequence data for the locus in question received from a data source may be broken into fragments, with each fragment containing one or more k-mers. These k-mers may be quickly compared to the list of k-mers in the locust database to determine whether the sequence data corresponds to the reference sequence or to a variation of the reference sequence.