Genome Data Analysis Device Using Burrows-Wheeler Transform for Variation Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current DNA sequencing technologies face high computational costs and bias in mapping processes when analyzing vast amounts of data from next-generation DNA sequencers, making it impractical to handle all read sequences simultaneously and accurately identify variations across the human genome.
Innovation Solution
A data analysis method that creates databases for both read sequences and reference genomes, using the Burrows-Wheeler Transform to efficiently search for variations by sorting sequences lexicographically and determining the minimum length for uniqueness, allowing for neutral processing and reduced computational costs by focusing on specific gene regions and sequencing errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If all read sequences are processed simultaneously through mapping to accurately identify variations, then measurement precision is improved, but productivity deteriorates due to the huge computational cost
Solution Approach 1:
The patent divides the genome into multiple non-overlapping blocks and processes read sequences in segments rather than all at once. Each block is processed independently through mapping and variant calling, allowing parallel computation while maintaining accuracy. This segmentation enables the system to handle large datasets by breaking them into manageable chunks that can be processed efficiently.
Solution Approach 2:
The patent performs preliminary actions by first identifying high-confidence variant positions through initial processing, then using these positions to guide subsequent detailed analysis. The system pre-processes data to identify regions of interest before performing comprehensive variant calling, reducing the overall computational burden while maintaining accuracy.
2Productivity
If mapping process is performed with heuristic selection, then productivity is improved by reducing computational cost, but object-affected harmful factors worsen due to bias in mapping results
Solution Approach 1:
The patent incorporates feedback mechanisms where mapping results are continuously evaluated and adjusted. The system uses iterative refinement where initial mapping results inform subsequent mapping parameters, and variant calling results feed back into improving mapping accuracy. This feedback loop reduces bias while maintaining computational efficiency.
Solution Approach 2:
The patent dynamically changes mapping parameters based on the specific characteristics of the data being processed. Different regions of the genome are processed with optimized parameters tailored to their specific properties, reducing bias while maintaining high productivity. The system adjusts mapping stringency, block sizes, and other parameters adaptively.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The invention can reduce the computational cost for a mapping process performed in analyzing genome/exome/transcriptome, and avoid generation of a heuristic-dependent bias, which would otherwise occur if a mapping destination is selected at one's discretion. To this end, the present invention sorts all cyclic permutations or suffixes of all read sequences to allow a search to be performed on the basis of any base sequence as a key (105). Additionally, the present invention computes and stores, for each base position in a genome sequence, the minimum length for uniqueness (MLU) at which a partial sequence starting from the base position becomes unique in the genome (110). In analysis of variations, a target region is scanned to inspect the number of matching read sequences that contain a partial sequence with a length of MLU and thus estimate the position of a variation (144), and then, the read sequences are collected at a position where the possibility of a variation having occurred is high to perform comparison analysis of the sequences (117).