Structural Variation Detection Using Barcode Overlap in Sequencing Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing algorithms for detecting structural variants and haplotype phasing are limited by their reliance on read pairs spanning breakpoints, making them ineffective for targeted sequencing data, such as whole exome sequencing, where breakpoints are often not spanned, and are computationally challenging due to the NP-hard nature of haplotype assembly.
Innovation Solution
The method involves fragmenting and partitioning the genome, using barcodes to identify structural variations by detecting regions with significant barcode overlap and obtaining phasing information, with a computer system processing sequence reads to determine the likelihood of structural variations by comparing barcode metrics to a threshold criterion.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing algorithms rely on read pairs spanning breakpoints for structural variant detection, then breakpoint detection accuracy is improved, but applicability to targeted sequencing data deteriorates
Solution Approach 1:
The patent introduces barcodes as an intermediary element that links sequence reads to their original genomic context. Instead of relying on read pairs to directly span breakpoints, the barcode system mediates the connection between reads and breakpoints through shared barcode identification. This allows structural variants to be detected through barcode co-occurrence patterns rather than requiring physical read pair spanning, thereby enabling applicability to targeted sequencing data while maintaining detection accuracy.
2Loss of information
If haplotype assembly is performed using conventional methods, then phasing information can be obtained, but computational complexity increases due to NP-hard nature
Solution Approach 1:
The patent segments the haplotype assembly problem into smaller, manageable units by utilizing barcodes to group reads into discrete families. Each barcode represents a segment of the genome with associated reads, and haplotype phasing is performed independently within each barcode family. This segmentation transforms the global NP-hard problem into multiple smaller subproblems that can be solved efficiently, reducing overall computational complexity while preserving phasing information.
Solution Approach 2:
The patent performs preliminary grouping of reads by barcode before attempting haplotype assembly. By pre-organizing reads into barcode-associated families and identifying shared barcodes across different genomic regions, the system prepares the data in a structure that facilitates efficient phasing. This preliminary action reduces the search space for haplotype assembly algorithms, making the otherwise NP-hard problem computationally tractable.
3Adaptability or versatility
If barcode-based methods are used for structural variant detection, then applicability to targeted sequencing data is improved, but detection of distant breakpoints deteriorates
Solution Approach 1:
The patent merges information from multiple barcode-associated read families to detect structural variants. By identifying barcodes that appear in multiple different genomic regions or bins, the system can infer the presence of structural variants even when the breakpoints are distant from each other. This merging of barcode information across different genomic locations extends the effective detection range beyond what single read pairs could achieve.
Data Source
AI summary
Systems and methods for determining structural variation and phasing using variant call data obtained from nucleic acid of a biological sample are provided. Sequence reads are obtained, each comprising a portion corresponding to a subset of the test nucleic acid and a portion encoding a barcode independent of the sequencing data. Bin information is obtained. Each bin represents a different portion of the sample nucleic acid. Each bin corresponds to a set of sequence reads in a plurality of sets of sequence reads formed from the sequence reads such that each sequence read in a respective set of sequence reads corresponds to a subset of the nucleic acid represented by the bin corresponding to the respective set. Binomial tests identify bin pairs having more sequence reads with the same barcode in common than expected by chance. Probabilistic models determine structural variation likelihood from the sequence reads of these bin pairs.