ctDNA Genome-Wide Integration for Low-Abundance Mutation Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for detecting low abundance somatic mutations in circulating tumor DNA (ctDNA) are limited by low sensitivity and accuracy, especially in early-stage cancer diagnosis, due to the scarcity of input material and high false positive rates, which undermines the effectiveness of existing sequencing technologies.
Innovation Solution
A machine learning-based approach using convolutional neural networks (CNN) and error suppression protocols to filter sequencing noise and discriminate between cancer-related mutations and sequencing errors, integrating genome-wide mutational information for improved detection and diagnosis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If ultra-deep sequencing is applied to detect low abundance somatic mutations in ctDNA, then detection sensitivity should improve, but the limited input material (hundreds to few thousands of genomic equivalents) renders sequencing ineffective and detection limit remains below 0.1-1% tumor fraction
Solution Approach 1:
The patent segments the detection approach by dividing the genome into targeted regions of interest and focusing sequencing efforts on these specific loci rather than attempting to sequence the entire genome at ultra-deep coverage. This allows sufficient sampling depth at critical positions while reducing the total material requirement.
Solution Approach 2:
The patent introduces molecular amplification and enrichment techniques as intermediaries between the limited input material and the sequencing process. These intermediaries amplify the signal from scarce tumor DNA fragments, enabling detection below the 0.1-1% tumor fraction threshold that would be impossible with direct sequencing.
2Reliability
If mutation callers like MUTECT are used to differentiate true mutations from sequencing errors, then false positive rate should decrease, but sensitivity drops to below 0.1 when mutation allele frequency is 0.05 and sequencing depth is 10x
Solution Approach 1:
The patent changes the parameter of sequencing depth from shallow (10x) to ultra-deep coverage at targeted positions, and adjusts the allele frequency threshold for mutation calling. By increasing the number of observed reads at each position, the patent enables reliable distinction between true low-frequency mutations and sequencing errors even at 0.05 allele frequency.
Solution Approach 2:
The patent implements dynamic mutation calling thresholds that adapt to the observed sequencing depth and quality metrics at each position. Rather than using fixed thresholds, the system dynamically adjusts sensitivity parameters based on the actual data quality and coverage, enabling optimal balance between false positive rate and detection sensitivity.
3Measurement precision
If CT scanning is used for early cancer detection, then detection coverage should improve, but false positive rate increases leading to costly and potentially risky follow-up evaluation
Solution Approach 1:
The patent replaces the mechanical imaging approach of CT scanning with a molecular-level detection system that directly analyzes tumor-specific DNA mutations in circulating blood. This substitution enables detection at the genetic level, providing higher specificity by identifying cancer-derived molecular signatures rather than relying on anatomical imaging that captures both malignant and benign findings.
Data Source
AI summary
The disclosure relates to systems, software and methods for diagnosing tumor diseases in a patient.