Flow Cell Image Registration for Sub-Pixel Cluster Base Calling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing algorithms for next-generation sequencing struggle with accurately identifying cluster centers in DNA sequencing due to precision issues and registration problems, leading to improper sequence identification and increased processing time.
Innovation Solution
Implement computational methods to improve image resolution beyond physical limits by identifying candidate cluster centers at a sub-pixel level, determining purities, and using dedicated processors and FPGAs for real-time processing to enhance accuracy and efficiency in base-calling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing image processing algorithms are used to identify cluster centers, then the system is simple to implement, but the measurement precision of cluster center locations deteriorates due to pixel-level resolution limits
Solution Approach 1:
The patent transitions from pixel-level (2D discrete) to sub-pixel (continuous coordinate) identification of cluster centers. By fitting mathematical models (e.g., Gaussian functions) to the intensity distribution across multiple pixels, the system determines cluster center locations with precision beyond the physical pixel grid, effectively adding a dimensional layer of computational refinement to the spatial measurement.
Solution Approach 2:
The patent replaces direct optical/mechanical resolution improvement with computational image processing. Instead of using higher-resolution imaging hardware, the system uses algorithms (e.g., centroid calculation, Gaussian fitting) to extract sub-pixel location information from existing images, substituting mechanical/optical enhancement with mathematical processing.
2Measurement precision
If computational methods are used to improve image resolution beyond physical limits, then the measurement precision improves, but the processing time increases
Solution Approach 1:
The patent performs computational image processing and cluster center identification during the initial flow cycles before the main sequencing data collection. By completing the template generation and base-calling location identification in advance, the system prepares the reference framework early, allowing subsequent sequencing cycles to proceed without repeated heavy processing, thus amortizing the computational time cost.
Solution Approach 2:
The patent creates a computational template (a digital copy or model of the flow cell layout and cluster positions) that can be reused across multiple sequencing cycles. This template serves as a reference that eliminates the need to re-process images from scratch for each cycle, significantly reducing redundant processing time while maintaining precision through the established sub-pixel measurement framework.
3Reliability
If accurate base-calling locations are identified, then the reliability of sequence identification improves, but the device complexity increases due to additional processing steps
Solution Approach 1:
The patent implements self-alignment and self-correction mechanisms where the system uses the identified cluster centers and purities to automatically adjust base-calling accuracy. The purity calculations for each candidate cluster center provide intrinsic quality metrics that guide the selection of reliable locations, allowing the system to self-regulate accuracy without external intervention or complex additional hardware.
4Measurement precision
If sub-pixel level processing is implemented, then the measurement precision improves, but the computational resources and storage requirements increase
Solution Approach 1:
The patent extracts only the essential sub-pixel location information and purity metrics from the full images, storing these refined measurements rather than the complete high-resolution image data. By separating and storing only the critical extracted parameters (cluster center coordinates with sub-pixel precision and associated purity values), the system maintains measurement precision while significantly reducing storage requirements compared to retaining all raw image data.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
Enhances the accuracy of sequence identification by accurately attributing optical signals to the correct DNA fragment, reducing processing time, and minimizing storage requirements, while maintaining or reducing the cost of sequencing systems.
Implementation Method 1
The flow cell is exposed to excitation light, exciting the labels and causing them to fluoresce. Because the cloned strands are clustered together, the fluorescent signal for any one fragment is amplified by the signal from its cloned counterparts, such that the fluorescence for a cluster may be recorded by an imager.
Data Source
AI summary
Image data analysis, particularly identifying cluster locations for performing base-calling in a digital flow cell image during DNA sequencing, is described. Each nucleic acid template molecule immobilized on a support may include an insert sequence and a sample index sequence. The sample index sequence may include a k-mer sequence. A sequencing system may conduct k cycles of sequencing reactions of the k-mer sequence before conducting one or more cycles of the insert sequence sequencing reactions and generate a first plurality of flow cell images. Pixel intensities may be determined for pixels of the first plurality of flow cell images. A base calling template may be determined and include base calling locations based on the pixel intensities and respective color purities of the pixel intensities. The base calling template may register a second plurality of flow cell images of the support in one or more cycles subsequent to the k cycles.