Neural Network Base Calling for Overlapping Sequencing Clusters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current nucleic acid sequencing technologies face limitations in resolving data from closely proximate or spatially overlapping clusters, leading to inefficiencies in throughput and computational resource usage, particularly in cluster-based sequencing methods.
Innovation Solution
Employing neural network-based systems and methods, including deep convolutional neural networks, to enhance the resolution of nucleic acid sequencing data by analyzing subpixel regions and generating cluster metadata, thereby improving the accuracy and efficiency of base calling.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image analysis methods are used to sequence nucleic acid clusters, then the system is simpler to implement, but the ability to resolve closely proximate or spatially overlapping clusters is insufficient
Solution Approach 1:
The patent replaces traditional mechanical/image processing analysis methods with deep neural network-based analysis. The system uses convolutional neural networks (CNNs) to automatically extract features from sequencing images and distinguish between overlapping clusters, substituting complex computational algorithms for simpler traditional image analysis while achieving superior resolution of closely proximate clusters
Solution Approach 2:
The patent changes the analytical parameters by using deep neural networks to process image data at multiple scales and feature levels. The CNN architecture transforms raw pixel data into hierarchical feature representations, enabling the system to resolve overlapping clusters by analyzing multiple parameter dimensions simultaneously rather than relying on single-parameter thresholding methods
2Reliability
If deep neural networks are employed to analyze sequencing images, then the resolution and quality of sequencing data improve, but the computational resource usage increases
Solution Approach 1:
The patent applies preliminary action by using the deep neural network to generate cluster metadata (such as cluster boundaries, intensities, and positions) before the actual sequencing analysis. This pre-processing step organizes and structures the image data in advance, reducing the computational burden during subsequent sequencing operations and enabling more efficient resource utilization while maintaining high data quality
Solution Approach 2:
The patent uses the deep neural network to create accurate metadata copies of cluster information from the original images. Instead of repeatedly analyzing the full-resolution images during sequencing, the system works with compressed metadata representations that capture essential cluster characteristics, significantly reducing computational resource requirements while preserving sequencing accuracy
3Productivity
If traditional methods are used for cluster analysis, then computational resources are conserved, but the throughput and quantity of sequencing data are limited
Solution Approach 1:
The patent replaces traditional sequential image analysis methods with parallel deep neural network processing. The CNN architecture enables simultaneous analysis of multiple clusters across the entire imaging array, dramatically increasing throughput by processing all clusters in parallel rather than sequentially, while the automated feature extraction reduces system complexity management
Solution Approach 2:
The patent implements universality by designing a deep neural network system that performs multiple functions: it identifies clusters, determines their positions, calculates intensities, resolves overlaps, and generates metadata all within a single integrated model. This multi-functional approach increases throughput by eliminating the need for separate analysis steps while managing complexity through unified architecture
Data Source
AI summary
A system, a method and a non-transitory computer readable storage medium for base calling are described. The base calling method includes processing through a neural network first image data comprising images of clusters and their surrounding background captured by a sequencing system for one or more sequencing cycles of a sequencing run. The base calling method further includes producing a base call for one or more of the clusters of the one or more sequencing cycles of the sequencing run.