Sequencing Data Read Realignment for Indel Calling Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current sequencing technologies face challenges in accurately calling insertions and deletions (indels) due to their lower occurrence rates, difficulty in mapping to correct genome locations, presence of repeat regions, and the inability of aligners to correctly identify variants at the ends of sequencing reads, primarily due to the lack of a two-sided context for variant calling.

Innovation Solution

A computer-implemented method for sequencing data read realignment that involves obtaining initial alignments, performing realignment processing to introduce candidate indels into a flattened aligned read, and selecting the best realignment based on prioritization criteria such as known indels, length, frequency, and location, to provide a more accurate alignment with minimized mismatches.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current aligners are used to map reads to the genome, then the mapping process is simple and fast, but the accuracy of indel calling is poor due to inability to correctly identify variants at the ends of sequencing reads

Engineering Contradiction:
Improveindel calling accuracyVSAvoidrealignment process complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary identification of candidate indels from the initial alignment before conducting the actual realignment. By pre-identifying potential indel locations and characteristics, the system prepares the data structure needed for accurate realignment, thereby improving indel calling accuracy while managing complexity through staged processing

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The realignment process is divided into distinct segments: identifying candidate indels, determining their characteristics (insertion/deletion, length, frequency), and then performing realignment only at those specific locations. This segmentation allows the system to focus computational resources on problematic regions, improving accuracy without requiring complete re-alignment of all reads

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If realignment processing is performed to improve indel calling accuracy, then the precision of variant detection is improved, but the processing time increases

Engineering Contradiction:
Improveindel calling accuracyVSAvoidrealignment processing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies realignment processing selectively only to reads that contain candidate indels, rather than performing universal realignment on all reads. By applying the time-consuming realignment operation only where necessary (partial action), the system maintains high indel calling accuracy for problematic regions while minimizing overall processing time

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The system applies different processing qualities to different regions: candidate indel regions receive full realignment processing with high computational resources, while other regions use the simpler initial alignment. This local differentiation of processing quality improves accuracy where needed while preserving processing efficiency elsewhere

Inventive Principle:
Principle #3Local quality

3Reliability

If all candidate indels are considered for realignment, then the completeness of variant detection is improved, but the false positive rate increases due to low-frequency variants

Engineering Contradiction:
Improveindel calling reliabilityVSAvoidfalse positives
Core Design Contradiction:
ReliabilityVSObject-generated harmful factors

Solution Approach 1:

The patent changes the frequency parameter threshold for considering candidate indels during realignment. By setting a minimum frequency threshold (e.g., requiring indels to appear in multiple reads), the system filters out low-frequency artifacts that would cause false positives while maintaining detection of true variants that appear with sufficient frequency across the cohort

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The system uses feedback from the initial alignment results to inform the realignment process. By analyzing the frequency and distribution of candidate indels from the initial mapping, the system can adjust which candidates proceed to realignment, using the initial results as feedback to guide subsequent processing and reduce false positives

Inventive Principle:
Principle #23Feedback

4Productivity

If the initial alignment is used directly without realignment, then the processing speed is fast, but the accuracy of indel variant calling is poor

Engineering Contradiction:
Improveprocessing speedVSAvoidindel calling accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent performs preliminary analysis of the initial alignment to identify candidate indels before conducting full realignment. This preliminary step allows the system to quickly process reads that don't contain indels (maintaining high speed) while preparing targeted realignment for only those reads that need it (improving accuracy)

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system applies realignment partially - only to reads identified as containing candidate indels - rather than universally to all reads. This partial application maintains high processing speed for the majority of reads while improving indel calling accuracy for the subset that requires it

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3542293B1Methods of sequencing data read realignment
Publication Date: 2023.12.27 ILLUMINA INC
  • EP3542293B1 patent drawingFigure 1A~1D
  • EP3542293B1 patent drawingFigure 2
  • EP3542293B1 patent drawingFigure 3

AI summary

Sequencing data read realignment. A method obtains from a sequence alignment dataset an initial alignment of a read sequence to a reference sequence and performs realignment processing on the initial alignment. The realignment processing includes identifying candidate indel(s) that include zero or more indels in the aligned read and zero or more indels aligned proximal to the aligned read as indicated by the sequence alignment dataset, creating a flattened aligned read based at least on removing from the aligned read any indels indicated by the initial alignment, and determining candidate realignment(s) of the read sequence to the reference sequence based on introducing, for each candidate realignment of the candidate realignment(s), a respective at least one candidate indel of the candidate indel(s) into the flattened aligned read. The method further provides the initial alignment or a selected candidate realignment of the candidate realignment(s) based on selection criteria.