Layered Suffix Tree Index for Cache-Efficient Genome Mapping

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional BWA-MEM algorithms require large memory resources due to the enormous size of the index generated from the human genome reference sequence, leading to significant memory access time and limiting processing speed in genome analysis.

Innovation Solution

A hierarchical index generation method that structures a suffix array into a suffix tree with a predetermined number of levels, splitting it into a supertree and subtrees, allowing data to be stored in cache memory for faster access.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a complete suffix array index is generated from the human genome reference sequence, then mapping accuracy is improved, but memory resource consumption increases significantly

Engineering Contradiction:
Improvemapping accuracyVSAvoidmemory resource consumption
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The patent divides the complete suffix array index into multiple levels (L0, L1, L2, etc.), where each level contains a portion of the index data. This segmentation allows the system to load only necessary index portions into cache memory during mapping operations, significantly reducing memory resource consumption while maintaining mapping accuracy through hierarchical data organization.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent implements a nested hierarchical structure where level L0 contains the root node, level L1 contains child nodes of L0, level L2 contains grandchildren nodes, and so on. This nesting allows progressive loading of index data from main memory into cache memory in a controlled manner, enabling accurate mapping while managing memory resources efficiently through selective data loading.

Inventive Principle:
Principle #7Nested doll (Nesting)

2Quantity of substance

If the index is stored in external memory (main memory), then sufficient storage capacity is achieved, but processing speed is limited by memory access time

Engineering Contradiction:
Improvestorage capacityVSAvoidprocessing speed
Core Design Contradiction:
Quantity of substanceVSProductivity

Solution Approach 1:

The patent segments the large index into hierarchical levels that can be selectively loaded into cache memory. During mapping operations, only the necessary portions of the index (specific levels and nodes) are loaded from main memory into cache, enabling fast access and processing while maintaining adequate storage capacity in main memory for the complete index structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces cache memory as an intermediary between main memory and the processing unit. The hierarchical index structure enables efficient use of cache memory by loading only necessary index portions, acting as a buffer that speeds up access to frequently used index data while maintaining connection to the complete index stored in main memory.

Inventive Principle:
Principle #24Intermediary (Mediator)

3Loss of information

If random access to main memory is performed for each query character, then complete data retrieval is achieved, but access time accumulates significantly

Engineering Contradiction:
Improvedata retrieval completenessVSAvoidaccess time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

The patent performs preliminary organization of the index into a hierarchical structure with levels L0, L1, L2, etc., where data is pre-grouped by prefixes and frequency. This preliminary organization enables the system to predict which index portions will be needed and load them into cache memory in advance, reducing the need for random access to main memory during actual mapping operations and significantly reducing access time while maintaining complete data retrieval capability.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4708300A1Computer program for index creation, computer program for mapping, computer device for index creation, computer device for mapping, and data structure of layered index
Publication Date: 2026.03.11 ADVANCED ACCELERATING SYST CO LTD
  • EP4708300A1 patent drawingFigure 1
  • EP4708300A1 patent drawingFigure 2
  • EP4708300A1 patent drawingFigure 3

AI summary

The present invention is an index generation computer program for implementing the generation of an index to be used in identifying a partial character string of a reference character string based on a query character string. The index generation computer program includes the step of setting a number of levels from an uppermost level to a lower level in a suffix tree, which is obtained by hierarchically structuring a suffix array of the reference character string, to a predetermined value and thereby omitting hierarchical data at levels lower than the predetermined value. With this configuration, analysis of given query data using the reference data can be performed at high speed and/or efficiently.