Sequence Assembly via Contig-Mediated Read Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for sequencing and genotyping face challenges in accurately assembling and aligning nucleic acid reads due to limitations in contig assembly algorithms, which lead to exclusion of legitimate reads and poor detection of mutations, especially indels and substitutions near read ends, and are computationally intractable for high-throughput data analysis.
Innovation Solution
Combining contig-based sequence assembly with individual read alignment to a reference genome, using techniques like Burrows-Wheeler Transform and barcode information to reduce computational complexity and improve positional accuracy, allowing for accurate detection of mutations by translating raw read alignments through contig alignments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If contig assembly algorithms are used to assemble reads into contigs and align to reference, then sequence assembly can be performed, but up to 10% of legitimate reads are excluded due to quality thresholds and the method is computationally intractable for high throughput data
Solution Approach 1:
The patent segments the read processing task into two independent parts: (1) assembling a subset of high-quality reads into contigs, and (2) aligning all reads (including previously excluded low-quality reads) to the assembled contigs. This segmentation allows each step to be optimized independently, improving both reliability of read inclusion and computational efficiency for high throughput data.
Solution Approach 2:
The patent introduces assembled contigs as an intermediary structure between the raw reads and the reference genome. Instead of directly comparing all reads to the reference (computationally intractable) or excluding low-quality reads (reducing reliability), the contigs serve as a mediator that captures sequence information while enabling efficient alignment of all reads through the contig-to-reference alignment results.
2Measurement precision
If alignment algorithms favor mismatch detection, then substitution sensitivity is improved, but indel sensitivity deteriorates, and vice versa
Solution Approach 1:
The patent employs dynamic adjustment of alignment parameters based on the specific analysis needs. By performing alignment to contigs first (which captures both mismatches and indels in the assembled sequence) and then using those results to inform subsequent read alignments, the system adaptively optimizes detection of different mutation types without being constrained by fixed algorithmic tradeoffs.
Solution Approach 2:
The patent maintains continuous improvement in mutation detection by using the contig assembly results as a foundation that informs and enhances subsequent read alignments. The information from contig-to-reference alignment continuously guides the interpretation of individual read alignments, ensuring both mismatch and indel detection capabilities are maintained and improved throughout the analysis pipeline.
3Measurement precision
If individual reads are aligned to reference genome, then positional accuracy can be obtained, but very short reads may align well in multiple places offering little positional accuracy and long indels are difficult to detect
Solution Approach 1:
The patent merges the information from two alignment levels: (1) contig alignment to reference genome, and (2) individual reads to contigs. By combining these alignment results, the system achieves both the positional accuracy of reference-based alignment and the reliability of unique mapping, since reads are positioned through the intermediate contig structure that has already been uniquely placed on the reference.
Solution Approach 2:
The patent implements a nested alignment structure where individual reads are aligned to contigs, which themselves are aligned to the reference genome. This nested approach allows short reads to inherit positional information from their parent contig, improving positional accuracy without requiring the reads themselves to uniquely map to the reference. The nested structure also helps detect long indels that span multiple reads within a contig.
Data Source
AI summary
The invention relates to assembly of sequence reads. The invention provides a method for identifying a mutation in a nucleic acid involving sequencing nucleic acid to generate a plurality of sequence reads. Reads are assembled to form a contig, which is aligned to a reference. Individual reads are aligned to the contig. Mutations are identified based on the alignments to the reference and to the contig.