GC Bias Correction in Noninvasive Prenatal Sequencing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Massively parallel sequencing of circulating fetal nucleic acids for noninvasive prenatal diagnosis is hindered by quantitative biases, particularly due to GC content, which affects the accuracy of read distributions and clinical diagnosis.
Innovation Solution
The methods involve improving alignment by allowing mismatches in index sequences and using non-repeat masked human reference genomes, and reducing GC bias through correction algorithms such as linear and LOESS regression, as well as modifying genomic representation calculations to enhance the precision of read counts and diagnostic accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If massively parallel sequencing is used for noninvasive prenatal diagnosis, then the ability to detect fetal chromosomal abnormalities is improved, but quantitative biases in sequencing data reduce measurement precision
Solution Approach 1:
The patent applies GC bias correction algorithms that adjust sequencing data based on the GC content parameter of genomic regions. By modeling and correcting the relationship between GC content and sequencing depth, the method transforms biased quantitative data into corrected estimates that accurately reflect true chromosomal representation, thereby resolving the measurement precision issue while maintaining detection reliability
Solution Approach 2:
The patent replaces direct physical measurement of chromosomal abundance with a computational modeling approach. Instead of relying on raw sequencing counts which are mechanically biased by GC content, the system uses statistical models to substitute and correct the measurement process, converting biased physical data into accurate biological information
2Speed
If repeat-masked reference genomes are used for alignment, then alignment speed is improved, but the number of uniquely aligned reads decreases
Solution Approach 1:
The patent segments the reference genome into repeat-masked and unmasked portions, using a two-stage alignment approach. First, reads are aligned to the repeat-masked reference for rapid initial mapping. Second, reads that fail to uniquely align are aligned to the unmasked reference to recover additional uniquely mapped reads. This segmentation resolves the contradiction by combining speed benefits with quantity recovery
Solution Approach 2:
The patent performs preliminary alignment to the repeat-masked reference genome before attempting alignment to the complete unmasked reference. This preliminary action quickly filters out most reads that can be confidently mapped, allowing the more computationally intensive unmasked alignment to focus only on ambiguous reads, thereby maintaining overall speed while increasing total uniquely aligned reads
Data Source
Figure 1
Figure 4.1
Figure 4.2
AI summary
This invention provides several ways of managing GC bias that occurs during sequencing and analysis of genomic DNA. Maternal plasma can be used as a source of fetal DNA for analysis. DNA segments or tags obtained from the plasma can be aligned with a chromosomal region of interest and with an artificial reference chromosome assembled from regions of the genome having matching GC content. This technology can be used, for example, to detect and evaluate aneuploidy and other chromosomal abnormalities