Sequence Variant Call Validation for False Positive Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sequencing techniques experience false positives in variant calling due to systematic errors such as FFPE artifacts, sequencing errors, or PCR errors, particularly when analyzing multiple types of DNA variants in a single sample.
Innovation Solution
A computer-implemented method and system for validating variant calls by analyzing sample and baseline variant frequencies to obtain a quality score, using a non-parametric Wilcoxon rank sum test and threshold comparisons to confirm or invalidate potential variant calls, and employing a hierarchy-based decision tree to filter calls.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If next-generation sequencing is used to assess variants in multiple genes using one sample, then the ability to detect multiple DNA alterations simultaneously is improved, but the number of false positive variant calls increases due to systematic errors
Solution Approach 1:
The patent applies preliminary action by establishing a baseline model of systematic errors from normal genomic sequences before analyzing tumor samples. The system pre-characterizes error patterns from FFPE artifacts, sequencing errors, and PCR errors using normal samples, then uses this baseline to correct and validate variant calls in tumor samples, reducing false positives while maintaining the ability to detect multiple alterations
Solution Approach 2:
The patent implements feedback by using the baseline error model to continuously validate and adjust variant call confidence scores. The system compares observed variant frequencies against the baseline model, feeds back the discrepancy information, and adjusts the reliability assessment of each variant call accordingly, allowing the system to distinguish true variants from systematic errors
2Adaptability or versatility
If variant calling is performed on FFPE samples with low abundance DNA, then the ability to analyze degraded samples is improved, but the accuracy of variant detection deteriorates due to artifacts and errors
Solution Approach 1:
The patent uses the baseline model derived from normal samples as an intermediary to mediate between the noisy FFPE sample data and the true variant signal. The baseline model acts as a reference framework that helps interpret variant frequencies in degraded samples, filtering out FFPE-specific artifacts while preserving genuine low-abundance variants
Solution Approach 2:
The patent applies parameter changes by adjusting the baseline model parameters based on the specific characteristics of each FFPE sample, such as DNA degradation level and coverage depth. The system dynamically modifies the baseline expectations to account for sample-specific artifacts, thereby maintaining measurement precision across varying sample qualities
Data Source
AI summary
Methods and systems are provided for validating variant calls. Sequencing data is received for a sample read along the genomic sequence of interest. An indication is received of a potential variant call at a designated position within the sequence of nucleotides along the genomic sequence of interest. The methods and systems obtain baseline variant frequencies at the designated position within one or more baseline genomic sequences, determine a sample variant frequency at the designated position for the genomic sequence of interest, analyze the baseline and sample variant frequencies at the designated position to obtain a quality score, and validate the potential variant call for the genomic sequence of interest based on the quality score.