Negative Information Storage Model for Genomic Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current genomic data storage strategies face challenges with vast storage needs and computational resource burdens due to storing all known information, leading to inefficiencies and increased risks of false negatives when handling large datasets.
Innovation Solution
The proposed method involves defining a 'positive region' for each sample, storing only genetic variants and missing data, and inferring reference states when neither are present, allowing for reduced storage and computational requirements by assuming missing data rates of 5-15%, thereby saving 85-95% of storage and computational resources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all known genomic information is stored, then complete data coverage is achieved, but storage requirements and computational resource burdens increase significantly
Solution Approach 1:
The patent extracts and stores only the essential information needed to represent the genomic data - specifically storing variants and missing data indicators - while inferring reference values rather than storing them explicitly. This extraction approach reduces storage requirements from billions of data points to only the deviations from reference, directly resolving the contradiction between complete data coverage and storage quantity.
Solution Approach 2:
Instead of storing what is known (reference values and all data), the patent inverts the approach by storing what is NOT known (variants and missing data). This inversion allows the system to achieve complete data coverage through inference while minimizing storage requirements, as the majority of genomic positions can be inferred as reference values when no variant or missing indicator is stored.
2Quantity of substance
If only variants are stored to reduce storage needs, then storage requirements decrease, but precision is lost and false negatives increase
Solution Approach 1:
The patent applies local quality by differentiating between three types of data states at each genomic position: reference values (inferred when no storage entry exists), variants (stored with specific indicators), and missing data (stored with specific indicators). This local differentiation ensures that each position is handled according to its actual state, maintaining precision while allowing reference positions to be inferred rather than stored, thus resolving the contradiction between storage reduction and accuracy maintenance.
3Quantity of substance
If reference values are assumed for all positions, then storage is minimized, but false negatives increase when different genomic tests are performed
Solution Approach 1:
The patent performs preliminary action by establishing a structured storage framework that explicitly accounts for three data states (reference, variant, missing) before any queries or analyses are performed. This preliminary structuring with unique keys and state indicators ensures that when different genomic tests are performed, the system can accurately determine whether a position should be interpreted as reference, variant, or missing, preventing false negatives while maintaining minimal storage requirements.
Data Source
AI summary
Systems and methods for storing large data sets, such as genetic sequence information. Within a “targeted subset” of positions with information, the system stores, both variant states and missing states at each position. Reference states are not stored, but are inferred within the targeted subset when neither a variant nor a missing state is stored at a given position. The absence of a variant state at a given position is assumed to be a reference state. The criteria for missing data are defined in pre-processing and are customizable based on the use case. For example, each data point may represent the genetic information of a sample at a position in the genome. The targeted subset may represent those positions that were included in a sequencing test.