Assisted Local Alignment Using Sequence Variation Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for aligning short sequence reads to large and complex reference genome data are computationally intensive and inefficient, particularly due to the complexity of algorithms like Smith-Waterman, which requires comparing large datasets, and struggle with the vast genetic variation within species, making it challenging to achieve accurate and fast alignment.
Innovation Solution
The implementation of assisted local alignment using a reference graph, such as a sequence variation graph (SVG), which indexes and segments data to focus alignment efforts on matching blocks and candidate regions, reducing the amount of data to be aligned by aligning only areas between these blocks, thereby decreasing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional alignment algorithms like Smith-Waterman are applied to align sequence reads to large reference genome data, then alignment accuracy is improved, but computational complexity and processing time increase significantly
Solution Approach 1:
The patent segments the reference genome data into multiple smaller reference sequences or regions. Instead of aligning reads against the entire large reference genome at once, the system divides the reference into manageable segments and performs alignment on each segment separately. This segmentation reduces the computational complexity of each alignment operation while maintaining overall alignment accuracy through comprehensive coverage of all reference segments.
Solution Approach 2:
The patent extracts and identifies matching blocks (seeds) between the sequence read and reference data, then uses these extracted matches to guide the alignment process. By extracting only the relevant matching regions and using them as anchors, the system avoids the need to computationally compare the entire read against the entire reference, thereby reducing computational complexity while preserving alignment precision in the critical matching regions.
2Reliability
If the entire reference genome data is used for alignment, then alignment completeness is improved, but the amount of data to be processed increases exponentially
Solution Approach 1:
The reference genome is segmented into smaller reference sequences or regions, allowing the system to process only relevant portions of the reference data for each read alignment operation. This segmentation maintains alignment completeness by ensuring all possible reference regions are covered through systematic segmentation, while dramatically reducing the quantity of data that needs to be processed in each computational step.
Solution Approach 2:
The patent performs preliminary actions by pre-processing the reference genome data to create indexed structures or pre-identifying potential matching regions. This preliminary organization of reference data allows for faster retrieval and processing of only the relevant portions during actual alignment operations, reducing the effective data quantity to be processed while maintaining comprehensive coverage of the reference genome.
3Productivity
If global search algorithms are used to find local regions for alignment, then search efficiency is improved, but the computational cost of the search algorithm offsets the reduction in search space
Solution Approach 1:
The patent creates simplified representations or copies of the reference genome data in the form of indexed structures or condensed reference sequences. These copies allow for rapid searching and filtering of relevant regions without requiring full processing of the entire reference genome. The indexed copies enable efficient retrieval of potential matching blocks while avoiding the computational cost of exhaustive search through the complete reference data.
Solution Approach 2:
The patent introduces intermediary data structures such as indexes, hashes, or pre-computed reference representations that mediate between the sequence read and the full reference genome. These intermediaries enable efficient searching and filtering of relevant reference regions without requiring direct comparison of the entire read against the complete reference, thereby improving search efficiency while reducing computational cost.
Data Source
AI summary
A method of aligning a data sequence to one or more reference sequences represented as a sequence variation graph (SVG) is disclosed. The method can comprise receiving one or more alignment candidate regions and corresponding ordered seeding information. For each of the received alignment candidate regions, a current seed is determined, the current seed being a next-in-order unprocessed seed based on the ordered seeding information. Data paths in the alignment candidate region are then traversed to identify potential next seeds relative to the current seed. If at least one potential next seed is found, a next seed is selected and alignment results are generated by applying a local alignment procedure to align query data in portions of the query data sequence between the current seed and the next seed with reference data in portions of the alignment candidate region located between the current seed and the next seed.