Directed Graph Mapping for Heterozygous Genome Assembly
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current genome assembly methods face challenges in handling complex genomes due to limitations in mapping long-read sequences, particularly in identifying structural variants and handling nonlinear topological components, which are essential for accurate assembly of diploid and polyploid genomes.
Innovation Solution
A method using a directed graph representing a heterozygous genome with nonlinear topological components to map long sequence reads, allowing for detection of structural and single nucleotide variants by forming alignments across branches and paths within the graph, enabling improved genome assembly and phasing of haplotigs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional mapping methods are used for long-read sequences, then mapping speed is maintained, but mapping accuracy and structural variant detection capability deteriorate
Solution Approach 1:
The patent segments the long-read mapping problem into multiple phases: initial seeding with short k-mers, extension to full read length, and graph-based alignment. This segmentation allows the system to maintain accuracy for long reads while managing computational complexity through staged processing.
Solution Approach 2:
The patent transitions from traditional linear reference genome mapping to graph-based mapping that incorporates multiple genomic paths and topological structures. This dimensional change from 1D linear to 2D/3D graph space enables simultaneous handling of multiple variants and structural complexities, improving mapping accuracy for diverse genomic regions.
2Productivity
If greedy alignment algorithms are used, then assembly speed is improved, but assembly accuracy and handling of homologous regions deteriorates
Solution Approach 1:
The patent implements dynamic adjustment of alignment parameters and strategies based on local genomic context. The system adapts its approach for different regions (e.g., repetitive vs. unique regions, high GC vs. low GC), switching between greedy and more rigorous alignment methods as needed, thus maintaining both speed and accuracy across diverse genomic landscapes.
Solution Approach 2:
The patent applies different alignment quality standards and parameters to different genomic regions. For example, more stringent alignment criteria are applied in repetitive or homologous regions where accuracy is critical, while faster greedy approaches are used in unique, low-complexity regions. This local quality differentiation resolves the contradiction between speed and accuracy.
3Adaptability or versatility
If short-read sequences are used, then mapping simplicity is maintained, but coverage of structural variations and repetitive regions deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-processing long reads through quality filtering, trimming, and initial alignment to identify potential mapping locations before full graph-based alignment. This preliminary processing reduces the computational burden of subsequent detailed alignment while ensuring that structurally variant and repetitive regions are properly captured.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Technical solutions for mapping long nucleic acid sequence reads to a target sequence are provided. A directed graph, representing all or some of a genome and comprising one or more nonlinear topological components, is obtained for an organism having a heterozygous genome. Each nonlinear topological component has an initiating node and a terminal node connected by at least a first branch and a second branch. One of these branches corresponds to the target sequence. The directed graph uses a plurality of sequence reads from a biological sample of the organism. The sequence reads are overlapped by an unrestricted overhang amount, provided there is a minimum consensus region between each two sequence reads. A query sequence, encompassing at least the initiating node or the terminal node of a first nonlinear topological component, is obtained. The directed graph is used to form a mapping of the query sequence to the directed graph.