Haplotype-Resolved Genome Assembly via Fused Graph Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current genome assembly methods struggle to accurately distinguish between homologous sequences on different chromosomes and repetitive regions, leading to fragmented assemblies and difficulties in reconstructing haplotype-specific sequences, especially in diploid and polyploid genomes.
Innovation Solution
The method generates a fused assembly graph from reads of both haplotypes, merges it with haplotype-specific assembly graphs, removes cross-phasing edges, and reconstructs haplotype-specific haplotigs to create a final haplotype-resolved assembly graph, integrating structural variations and phased SNPs for comprehensive haplotype sequence reconstruction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional genome assembly methods are used, then assembly process is simple, but assembly accuracy deteriorates due to inability to distinguish homologous sequences from repetitive regions
Solution Approach 1:
The patent segments the genome assembly process into distinct phases: constructing a fused assembly graph from all reads, then separating it into haplotype-specific graphs using phased reads. This segmentation allows accurate distinction between homologous sequences and repetitive regions by processing each haplotype independently while maintaining their relationships.
Solution Approach 2:
The patent introduces phased reads as an intermediary element that mediates between the fused assembly graph and final haplotype resolution. These phased reads serve as the connecting link that enables accurate separation of homologous chromosomes by providing phase information about allele relationships.
2Measurement precision
If simple string matching algorithms are used, then processing speed is fast, but variant identification accuracy deteriorates due to sequencing errors
Solution Approach 1:
The patent performs preliminary actions by constructing a fused assembly graph that pre-processes all sequencing reads before variant identification. This pre-assembly step organizes the data structure to facilitate more accurate and efficient subsequent processing, resolving overlaps and establishing contig relationships ahead of time.
Solution Approach 2:
The patent replaces simple mechanical string matching algorithms with a sophisticated graph-based computational system. The fused assembly graph structure enables accurate variant identification by modeling complex relationships between reads, overlaps, and contigs, while the graph traversal algorithms provide efficient processing despite the increased complexity.
3Measurement precision
If haplotype-specific assembly graphs are constructed separately, then haplotype resolution is accurate, but assembly process becomes complex and time-consuming
Solution Approach 1:
The patent performs preliminary construction of a fused assembly graph that incorporates all reads before separating into haplotype-specific graphs. This pre-processing step establishes the overall genome structure and relationships, so that subsequent haplotype separation can proceed more efficiently with less redundant computation.
Solution Approach 2:
The patent merges the construction of fused and haplotype-specific assembly graphs into an integrated workflow. By constructing the fused graph first and then deriving haplotype-specific graphs from it using phased reads, the method combines the benefits of both approaches while reducing total processing time through shared computational steps.
4Quantity of substance
If reads from both haplotypes are processed together, then assembly completeness is improved, but ability to distinguish homologous sequences deteriorates
Solution Approach 1:
The patent segments the processed reads into haplotype-specific groups using phased reads after constructing the initial fused assembly graph. This segmentation maintains the completeness of using all reads while restoring the ability to distinguish homologous sequences by analyzing each haplotype's reads in context of its specific alleles and variants.
Solution Approach 2:
The patent applies local quality by treating different regions of the assembly graph with different processing strategies. In regions where phased reads provide strong haplotype signals, the method applies strict haplotype-specific filtering, while in regions with weaker signals, it maintains more permissive criteria to preserve assembly completeness.
Data Source
Figure 1
Figure 2
Figure 3A
AI summary
Exemplary embodiments provide methods and systems for diploid genome assembly and haplotype sequence reconstruction. Aspects of the exemplary embodiment include generating a fused assembly graph from reads of both haplotypes, the fused assembly graph including identified primary contigs and associated contigs; generating haplotype-specific assembly graphs using phased reads and haplotype aware overlapping of the phased reads; merging the fused assembly graph and haplotype- specific assembly graphs to generate a merged assembly haplotype graph; removing cross-phasing edges from the merged assembly haplotype graph to generate a final haplotype-resolved assembly graph; and reconstructing haplotype-specific contigs from the final haplotype-resolved assembly graph resulting in haplotype-specific contigs.