Non-acyclic Genome Graph Alignment on Heterogeneous Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sequence-to-graph alignment methods struggle with efficiency and accuracy when handling non-acyclic genome graphs, which are necessary for representing inter-individual and intra-individual variability in genomic data.
Innovation Solution
The method employs heterogeneous computing resources (CPUs and GPUs) to align sequence reads with non-acyclic sequence-labelled genome graphs, using a dual representation of the graph to enable intra-sequence parallelism and an adaptive bandwidth heuristic to optimize computation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If non-acyclic genome graphs are used to represent genomic variability, then alignment accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent segments the non-acyclic genome graph into multiple acyclic subgraphs (DAGs) by identifying and removing cyclic regions. Each subgraph can be processed independently using efficient DAG-based alignment algorithms, thereby maintaining alignment accuracy while reducing computational complexity. The segmentation allows the system to handle complex genomic variations without the full computational burden of processing the entire non-acyclic graph at once.
Solution Approach 2:
The patent performs preliminary actions by pre-processing the genome graph to identify cyclic regions and prepare acyclic subgraphs before the actual alignment process. This includes building indexes for the acyclic portions and pre-computing alignment parameters, which reduces the computational load during the main alignment phase and improves overall processing efficiency while maintaining accuracy.
2Manufacturing precision
If fine-grained non-acyclic graphs are used, then alignment precision is improved, but execution time increases significantly
Solution Approach 1:
The patent divides the fine-grained non-acyclic graph into multiple acyclic subgraphs, allowing parallel processing of alignment tasks. Each subgraph can be aligned independently and simultaneously, which maintains the high precision benefits of fine-grained representation while significantly reducing total execution time through parallel computation.
Solution Approach 2:
The patent applies partial action by focusing computational resources on aligning reads to specific acyclic subgraphs that are most relevant to the query sequences. Rather than exhaustively processing the entire graph with equal intensity, the method selectively processes portions of the graph that provide the most value for alignment precision, thereby reducing overall execution time while maintaining high precision where needed.
3Reliability
If exact alignment algorithms are used for non-acyclic graphs, then alignment quality is improved, but memory cost and execution time increase
Solution Approach 1:
The patent segments the genome graph into acyclic subgraphs, which can be processed using memory-efficient DAG-based alignment algorithms. This segmentation allows exact alignment to be performed on smaller, manageable portions of the graph, reducing the peak memory requirements compared to processing the entire non-acyclic graph with exact alignment algorithms.
Solution Approach 2:
The patent creates simplified copies or representations of the acyclic subgraphs that are optimized for memory-efficient processing. These copies contain only the necessary information for alignment tasks, allowing exact alignment algorithms to run with reduced memory footprints while maintaining alignment quality.
4Productivity
If DAG-ification is applied to non-acyclic graphs, then processing efficiency is improved, but alignment quality degrades
Solution Approach 1:
The patent segments the non-acyclic graph into acyclic subgraphs that preserve the original graph's structural information and alignment properties. By carefully identifying boundaries between cyclic and acyclic regions, the segmentation maintains alignment quality at the subgraph level while enabling efficient DAG-based processing, thus avoiding the quality degradation associated with forced DAG-ification of the entire graph.
Solution Approach 2:
The patent applies different processing strategies to different parts of the graph: acyclic subgraphs are processed using efficient DAG-based algorithms, while cyclic regions are handled with specialized approaches that preserve alignment quality. This local quality approach ensures that each region is processed with the most appropriate method, maintaining overall alignment quality while improving processing efficiency.
Data Source
AI summary
The disclosed embodiments concern methods, apparatus, systems, and computer program products for computing the best alignment score between a set of query sequences and a non-acyclic sequence-labelled reference genome graph comprising a processor and non-transitory memory. In an embodiment, said computing is performed employing an alignment algorithm suitable for non-acyclic graph that leverages inter-sequence and intra-sequence parallelism, executed on a heterogeneous computing system, containing at least a CPU (Central Processing Unit) and a GPU (Graphics Processing Units).


