Artefact Variant Scoring in High Throughput Sequencing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for reducing artefact variants in high-throughput sequencing, such as those used in whole exome sequencing, rely heavily on experimenter expertise and are inefficient, often missing artefacts due to subjective judgments and limited filtration methods.
Innovation Solution
A method that automatically identifies and excludes artefact variants by building a population variant frequency database, scoring variant sites using a weighted formula, and dividing sequences into predetermined window sizes to analyze and exclude artefact variants based on total scores within those windows.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated scoring methods are implemented, then productivity and standardization are improved, but device complexity increases
Solution Approach 1:
The filtering process is divided into multiple independent scoring components (Score1 for variant type, Score2 for coordinate position, Score3 for LOF confidence, Score4 for population frequency). Each component evaluates a specific aspect of the variant independently, then their weighted sum produces the final Svar score. This segmentation enables automated processing while maintaining manageable complexity through modular design.
Solution Approach 2:
The method transforms qualitative expert judgment into quantitative parameters by assigning numerical scores to different variant characteristics. The weighted scoring system converts multiple qualitative factors (variant type, position, functional impact, population frequency) into a single quantitative metric (Svar), enabling automated comparison and filtering without requiring expert intervention.
2Reliability
If comprehensive filtration is applied, then reliability is improved, but loss of information increases due to potential false exclusions
Solution Approach 1:
The filtering approach applies different weightings and scoring criteria to different aspects of variant evaluation locally. For example, Score1 differentiates between variant types (SNV, INDEL, MNV) with different weights, Score2 applies position-specific penalties for artefact-prone regions, and Score3 applies LOF-specific confidence thresholds. This localized evaluation ensures that each variant is judged by criteria most relevant to its specific characteristics, improving reliability while minimizing false exclusions.
Solution Approach 2:
The method incorporates population frequency (Score4) as feedback from large-scale sequencing data to inform individual variant evaluation. By comparing each variant against population-level observations, the system can distinguish between true rare variants and artefacts that appear frequently across samples. This feedback mechanism improves artefact detection while preserving true variants that genuinely occur at low frequencies.
Data Source
AI summary
A method of reducing an artefact variant in high-throughput sequencing, which belongs to the technical field of bioinformatics is provided. The method includes the steps of: firstly building a population variant frequency database and integrating to obtain a population variant frequency at each variant site, along with total population, scoring each variant site using the formula designed by the inventors, and dividing a sequence into a predetermined window size after each variant to obtain a plurality of region sequences, then synthetically analyzing the scores of variant sites within each region sequence, and finally excluding the artefact variant sites based on identification. As a result, the method can achieve the detection of all types of artefact variants, and offers advantages such as high efficiency, automation, accuracy and comprehensive detection.