MPEG-G Genomic Data Compression via Configurable ML Predictors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The current MPEG-G standard for genomic data compression is limited by the lack of support for multiple contexts, complex predictors like neural networks, and adaptive modeling, leading to suboptimal compression ratios and inefficiencies in handling diverse genomic data types.
Innovation Solution
A framework is introduced that enhances MPEG-G by allowing multiple contexts, supports neural networks and machine learning models, and includes static, semi-adaptive, and adaptive training modes, enabling improved compression through flexible predictor selection and context adaptation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the current MPEG-G standard with CABAC is used for genomic data compression, then the implementation is simple and compatible with existing standards, but the compression ratio is suboptimal and cannot handle diverse genomic data types effectively
Solution Approach 1:
The patent implements dynamic model selection where the encoder can switch between multiple probability models (context models) based on the data being encoded. This allows the system to adapt to different genomic data patterns dynamically, improving compression ratio while maintaining implementation feasibility through a structured model selection framework.
Solution Approach 2:
The patent changes the parameter of context modeling by introducing multiple context models with different orders and types (e.g., quality value contexts, base context, position context). This allows the system to capture different statistical patterns in genomic data, significantly improving compression efficiency over the single-model CABAC approach.
2Productivity
If multiple contexts and machine learning predictors are introduced to improve compression efficiency, then the compression ratio and adaptability improve, but the device complexity and computational requirements increase
Solution Approach 1:
The patent segments the probability modeling into multiple independent context models, each specialized for different types of genomic data patterns. This segmentation allows the system to use simple, targeted models for specific data types rather than one complex universal model, improving compression efficiency while managing complexity through modular design.
Solution Approach 2:
The patent creates a universal framework that can handle multiple genomic data types (quality values, base calls, positions) using a common architecture with configurable context models. This multi-functionality allows the same encoder structure to adapt to different data types through parameter configuration rather than requiring separate complex encoders for each data type.
3Adaptability or versatility
If context-adaptive modeling with multiple contexts is implemented, then the ability to handle diverse genomic data types improves, but the computational overhead and processing time increase
Solution Approach 1:
The patent performs preliminary action by pre-defining multiple context models with their probability distributions before encoding actual genomic data. During encoding, the system quickly selects the appropriate pre-configured model based on data type indicators, avoiding the need to compute complex models in real-time and thus reducing processing time while maintaining high adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and a system for decoding MPEG-G encoded data of genomic information, including: receiving MPEG-G encoded data; extracting encoding parameters; selecting an arithmetic decoding type based upon the extracted encoding parameters; selecting a predictor type specifying the method to obtain probabilities of symbols which were used for arithmetically encoding the data, based upon the extracted encoding parameters; selecting arithmetic coding contexts based upon the extracted encoding parameters; and decoding the encoded data using the selected predictor and the selected arithmetic coding contexts.