Reduced Memory Nucleotide Sequence Comparison via Matrix Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for performing Smith Waterman analysis on nucleotide sequences require significant memory to store two-dimensional matrices, limiting parallel comparisons and efficiency due to slower access to external memory compared to on-chip memory.
Innovation Solution
Implementing a customized integrated circuit that stores and references only individual portions of the two-dimensional matrix, regenerating subsections as needed, and using reduced binary metadata for backtracking, allowing memory-efficient processing and parallel comparisons.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the traditional Smith Waterman algorithm stores the complete two-dimensional matrix in memory, then the alignment accuracy is maintained, but the memory consumption increases to several hundred kilobytes
Solution Approach 1:
The patent divides the large two-dimensional matrix into multiple smaller sub-matrices or blocks that can be stored and processed separately. This segmentation allows the system to maintain alignment accuracy by processing one block at a time while reducing overall memory consumption by not holding the entire matrix in memory simultaneously.
Solution Approach 2:
The patent introduces a temporal dimension by using iterative or sequential processing of matrix blocks. Instead of storing all spatial dimensions of the matrix at once, the system processes different blocks in sequence across time, effectively trading space for time while maintaining the ability to compute accurate alignments.
2Quantity of substance
If external memory is used to store the two-dimensional matrix, then memory capacity is sufficient, but the access speed becomes substantially slower compared to on-chip memory
Solution Approach 1:
By segmenting the matrix into smaller blocks that fit within on-chip memory capacity, the system can utilize the high-speed on-chip memory for active computation while only storing essential data structures in external memory, thereby improving access speed for the frequently accessed data.
Solution Approach 2:
The system performs preliminary processing to identify and store only the essential backtracking metadata and checkpoint information in on-chip memory before the actual backtracking process begins. This preliminary action ensures that the most frequently accessed data is already in fast memory, reducing access delays during the main computation phase.
3Ease of operation
If the complete two-dimensional matrix is stored in memory, then backtracking can proceed without regeneration, but the number of parallel sequence comparisons is limited due to memory constraints
Solution Approach 1:
The patent segments the backtracking process to operate on individual matrix blocks rather than the complete matrix. This allows multiple independent backtracking operations to proceed in parallel on different blocks, significantly increasing throughput while maintaining the integrity of each alignment computation.
Solution Approach 2:
The system creates and maintains copies of essential backtracking metadata and checkpoint information in on-chip memory for each parallel processing unit. This copying strategy enables multiple parallel comparisons to access their required data locally without interfering with each other, thereby increasing parallel throughput while keeping memory usage manageable.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Comparisons between two nucleotide sequences can be performed by customized integrated circuity that can implement a Smith Waterman analysis in a reduced memory footprint, storing and referencing only individual portions, or subsections, of a two-dimensional matrix that is representative of the comparison between the two nucleotide sequences. As the backtracking proceeds, backtracking metadata corresponding to a cell from a subsection that is not currently retained in memory can be required. Such a subsection can be regenerated from previously generated scores associated with checkpoint cells of the two-dimensional matrix that comprise two edges of the subsection being regenerated. Moreover, to further reduce memory consumption, the backtracking metadata stored for each cell can comprise four binary digits: two indicative of a directional assignment, one indicative of whether the corresponding cell is part of a deletion stretching across multiple contiguous cells, and one analogously indicative of insertions stretching across multiple contiguous cells.