HRD Genome Detection Using Shallow WGS and Gene Panel ML
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting Homologous Recombination Deficiency (HRD) in cancer genomes are time-consuming and costly, requiring high coverage sequencing and large training datasets, or rely on deep learning models that are difficult to train with limited data availability.
Innovation Solution
A method using Machine Learning (ML) that combines shallow Whole Genome Sequencing (WGS) data with non-shallow sequencing data from specific gene panels to determine HRD status, utilizing metrics such as Large-scale Genomic Alterations, Loss of Heterozygosity, and Copy Number Variations, and training an ML model to predict HRD presence.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high coverage sequencing with custom hybridization capture panel is used to determine HRD status, then measurement precision is improved, but loss of time and cost increase
Solution Approach 1:
The patent extracts only the essential genetic markers needed for HRD detection from the entire genome, focusing on specific panels of genes and genomic regions. This extraction approach maintains detection precision while eliminating the need for comprehensive high-coverage sequencing, thereby reducing time and cost without sacrificing accuracy
Solution Approach 2:
Instead of performing complete high-coverage sequencing of the entire genome, the patent applies partial sequencing by targeting specific gene panels and genomic regions that are most informative for HRD detection. This partial action approach achieves sufficient measurement precision for clinical decision-making while dramatically reducing the time and computational resources required
2Reliability
If deep learning models are trained on large datasets to evaluate HRD probability, then reliability is improved, but device complexity and data requirements increase
Solution Approach 1:
The patent uses a simplified machine learning model that requires only a modest training dataset rather than the large datasets needed by deep learning models. This partial approach achieves sufficient reliability for clinical HRD detection while dramatically reducing computational complexity, training data requirements, and infrastructure needs
Solution Approach 2:
The patent employs a lightweight machine learning model that can be trained and deployed with minimal computational resources, making it accessible in settings where expensive deep learning infrastructure is unavailable. This approach trades some theoretical maximum accuracy for practical accessibility and ease of deployment
3Loss of time
If shallow WGS data alone is used for HRD detection, then loss of time and cost are reduced, but measurement precision deteriorates
Solution Approach 1:
The patent merges shallow WGS data with targeted sequencing data from specific gene panels to achieve a synergistic effect. The shallow WGS provides broad genomic context while the targeted panels provide deep coverage of critical regions, together achieving higher precision than either approach alone while maintaining cost and time efficiency
Data Source
AI summary
A device for determining a HRD index of presence of Homologous Recombination Deficiency, HRD, in a genome of a subject, the device being configured for: receiving shallow WGS data and non-shallow sequencing data relative to a group of genes in the subject genome, obtaining at least one first parameter from the shallow WGS data and at least one second parameter from non-shallow sequencing data, and determining, by applying a HRD prediction Machine Learning Model to the obtained at least first and second parameters, an HRD index representative of presence of HRD in the subject genome.