Deep Neural Network for Sequencing Quality Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current DNA sequencing-by-synthesis techniques, such as SBS, are inefficient as they cannot predict the final quality of a sequencing run early on, leading to prolonged runs that often fail due to quality issues, necessitating the ability to halt poor runs quickly.
Innovation Solution
A deep learning-based system using quality prediction convolutional neural networks that analyze subsystem performance data to forecast the overall base calling quality during sequencing, allowing for early identification of potential run failures and reducing the number of necessary cycles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If sequencing-by-synthesis is performed to determine nucleic acid sequences, then sequence data can be obtained, but the run time becomes excessively long (three days or more) and quality cannot be predicted early
Solution Approach 1:
The system performs preliminary analysis of subsystem performance metrics during early cycles to predict final base calling quality before the run completes. By analyzing performance data from the first few cycles and using trained neural networks, the system can forecast the outcome and allow operators to halt poor runs early, preventing waste of time and resources on failed sequencing attempts.
2Reliability
If sequencing runs are completed to ensure quality results, then reliable sequence data is obtained, but runs that will fail continue for the full duration wasting resources
Solution Approach 1:
The system implements feedback by continuously monitoring subsystem performance metrics during the sequencing run and comparing them against patterns learned from historical data. The neural network processes real-time performance data and provides feedback predictions about final quality, enabling operators to make informed decisions about whether to continue or halt the run, thus optimizing resource utilization while maintaining reliability.
3Measurement precision
If operators wait for complete sequencing runs to assess quality, then final base calling quality is known, but time is lost and failed runs cannot be stopped early
Solution Approach 1:
The system performs preliminary quality assessment by analyzing subsystem performance metrics during early cycles and using trained neural networks to predict final base calling quality. This preliminary action provides quality estimates within hours rather than days, allowing operators to make timely decisions about run continuation without waiting for complete sequencing runs.
Data Source
AI summary
The technology disclosed predicts quality of base calling during an extended optical base calling process. The base calling process includes pre-prediction base calling process cycles and at least two times as many post-prediction base calling process cycles as pre-prediction cycles. A plurality of time series from the pre-prediction base calling process cycles is given as input to a trained convolutional neural network. The convolutional neural network determines from the pre-prediction base calling process cycles, a likely overall base calling quality expected after post-prediction base calling process cycles. When the base calling process includes a sequence of paired reads, the overall base calling quality time series of the first read is also given as an additional input to the convolutional neural network to determine the likely overall base calling quality after post-prediction cycles of the second read.