HRM Curve Classification via Neural Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for classifying high-resolution melt (HRM) curves are limited in their ability to probabilistically assess classification results and cannot effectively evaluate or classify curves across multiple thermal cycler runs or platforms, leading to inaccurate genotype classification.
Innovation Solution
Representing HRM curves as mathematical functions with varying coefficient values, using curve-fitting techniques like Chebyshev polynomial expansions, and employing artificial neural networks for classification, allowing for robust classification across multiple runs and platforms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If current software methods normalize HRM curves and use subtraction plots comparing with pre-defined controls, then the classification process is simple, but the method cannot probabilistically assess classification results and cannot evaluate curves across multiple runs or platforms
Solution Approach 1:
The patent transforms HRM curves from raw data into mathematical function representations with coefficients as parameters. This parameter transformation enables the curves to be processed by classification tools that can handle multiple runs and platforms, providing probabilistic assessment capabilities while maintaining systematic analysis.
Solution Approach 2:
The patent replaces the mechanical subtraction plot method with a neural network-based classification system. This substitution introduces probabilistic assessment capabilities and cross-platform evaluation while systematically processing the mathematical function coefficients representing HRM curves.
2Measurement precision
If HRM curves are classified using subtraction plots with pre-defined controls, then the method is easy to operate, but it assigns only yes/no values and cannot provide probabilistic assessment
Solution Approach 1:
By converting HRM curves into mathematical function coefficients, the patent enables the application of probabilistic classification algorithms. This parameter transformation allows the system to output probability values rather than simple yes/no classifications, improving measurement precision while maintaining operational simplicity through automated processing.
Solution Approach 2:
The patent creates a mathematical representation (copy) of the HRM curves using function coefficients. This copied representation can be processed by classification tools to generate probabilistic assessments, providing more precise classification results without requiring complex manual operations.
3Adaptability or versatility
If classification is limited to single thermal cycler usage, then the system is simple to implement, but it cannot evaluate and classify curves across multiple runs or platforms
Solution Approach 1:
The patent creates a universal classification system that processes mathematical function coefficients representing HRM curves. This universal approach enables the same classification tool to handle curves from multiple thermal cycler platforms and runs, providing platform adaptability while maintaining systematic processing through standardized mathematical representations.
Data Source
AI summary
The present invention relates to a method and system for classifying high-resolution melt (“HRM”) curves, and, more specifically, to a method and system for classifying HRM curves by genotype where the curves are represented by a mathematical function with varying coefficient values.