Dilated Convolutional Neural Network for DNA Basecalling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Legacy basecallers in capillary electrophoresis DNA sequencing face challenges with low accuracy for mixed bases, especially at 5′ and 3′ ends, and struggle with shorter amplicons, leading to increased error rates and difficulties in identifying variants like SNPs and het indels, due to sequencing artefacts and mobility shifts.
Innovation Solution
The implementation of convolutional neural network-based systems, specifically dilated convolutional neural networks, for basecalling in capillary electrophoresis, which outperform recurrent networks by achieving state-of-the-art accuracy and faster training and inference, enabling improved basecalling accuracy for pure and mixed bases through efficient scan labelling and quality value determination.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If recurrent neural network-based models are used for basecalling, then the ability to properly model time-series data is improved, but the processing speed is severely restricted
Solution Approach 1:
The patent replaces recurrent neural networks with convolutional neural networks, substituting a sequential processing mechanism with a parallel processing mechanism. This allows the system to maintain high basecalling accuracy while dramatically improving processing speed, as convolutional networks can process multiple time points simultaneously rather than sequentially waiting for earlier computations.
2Length of moving object
If legacy basecallers are used, then the system can handle long plasmid clones, but the accuracy for mixed bases and shorter amplicons deteriorates
Solution Approach 1:
The patent changes the fundamental parameters of the basecalling algorithm by transitioning from legacy basecalling methods to deep learning-based convolutional neural networks. This parameter change enables the system to accurately process both long reads and short amplicons, as well as correctly identify mixed bases, by learning complex patterns in the electropherogram data that traditional algorithms cannot capture.
3Measurement precision
If deep learning models are implemented, then basecalling accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the basecalling problem into distinct computational components handled by different parts of the convolutional neural network architecture. This segmentation allows the complex deep learning model to be broken down into manageable convolutional layers, pooling layers, and fully connected layers, each performing specific functions that collectively achieve high accuracy without overwhelming computational complexity.
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
This approach enhances the fidelity of Sanger sequencing data, improves variant identification, increases read length, and reduces sequencing costs by providing higher accuracy and faster processing for DNA sequencing applications.
Implementation Method 1
In capillary electrophoresis (CE), a biological sample, such as a nucleic acid sample, is injected at the inlet end of the capillary, into a denaturing separation medium in the capillary, and an electric field is applied to the capillary ends. The different nucleic acid components in a sample, e.g., a polymerase chain reaction (PCR) mixture or other sample, migrate to the detector point with different velocities due to differences in their electrophoretic properties.
Implementation Method 2
The magnitude of any given peak, including an artifact peak, is most often determined optically on the basis of either UV absorption by nucleic acids, e.g., DNA, or by fluorescence emission from one or more labelled dyes associated with the nucleic acid.
Data Source
AI summary
A method of automatically sequencing or basecalling one or more DNA (deoxyribonucleic acid) molecules of a biological sample is described. The method comprises using a capillary electrophoresis genetic analyzer to measure the biological sample to obtain at least one input trace comprising digital data corresponding to fluorescence values for a plurality of scans. Scan labelling probabilities for the plurality of scans are generated using a trained artificial neural network comprising a plurality of layers including convolutional layers. A basecall sequence comprising a plurality of basecalls for the one or more DNA molecules based on the scan labelling probabilities for the plurality of scans is determined.