Network Modeling for Gene Variant Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for identifying causative genes and genetic variants in diseased individuals are limited by the difficulty in prioritizing millions of genetic variants, especially in inherited diseases, due to statistical and computational limitations in traditional differential expression experiments, which restrict the scope of genome interpretation pipelines.
Innovation Solution
A method using network modeling to create Bayesian networks of causal interactions between genes, prioritizing genetic variants by identifying modular sub-networks and annotating variants in coding and non-coding regions, leveraging gene coexpression data and protein interaction data to infer relationships and predict pathogenicity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional differential expression experiments are used to identify disease genes, then the methodology is simple and straightforward, but the statistical and computational limitations restrict the scope of genome interpretation and reduce the ability to prioritize genetic variants
Solution Approach 1:
The patent introduces network models as intermediary structures that connect genetic variants to disease phenotypes through gene coexpression relationships. These network models serve as mediators that translate raw genomic data into prioritized candidate genes, overcoming the limitations of traditional direct association methods while managing computational complexity through structured relationships.
Solution Approach 2:
The patent transitions from one-dimensional gene-by-gene analysis to multi-dimensional network analysis by incorporating coexpression data, protein interactions, and pathway information. This dimensional expansion allows simultaneous consideration of multiple genetic variants and their relationships, greatly enhancing genome interpretation scope.
2Reliability
If the search space of candidate genes is restricted to known disease genes, then the analysis is more focused and manageable, but the ability to discover novel disease genes is reduced
Solution Approach 1:
The patent performs preliminary enrichment analysis using known disease genes and their network neighborhoods to establish baseline expectations. This preliminary action creates a reference framework that guides subsequent discovery while maintaining focus on biologically relevant candidates, balancing reliability with novelty.
Solution Approach 2:
The patent segments the genome into functional modules and network neighborhoods around known disease genes. By analyzing these segmented regions separately, the method can identify novel genes within functionally coherent units, improving both the reliability of discoveries and the manageability of analysis.
3Quantity of substance
If whole genome sequencing is performed to identify all genetic variants, then the comprehensiveness of variant detection is maximized, but the difficulty of prioritizing millions of variants increases
Solution Approach 1:
The patent extracts and prioritizes a small subset of high-confidence candidate genes from the millions of variants detected by whole genome sequencing. By focusing computational resources on network-connected genes with strong statistical evidence, the method makes the vast amount of sequencing data tractable while maintaining comprehensive variant detection.
Solution Approach 2:
The patent replaces manual or simple filtering approaches with automated network-based prioritization algorithms. This substitution enables systematic evaluation of millions of variants through computational network analysis, making the prioritization process scalable and objective.
Data Source
AI summary
With the advent of low cost, high-throughput whole genome sequencing (“next generation sequencing”), tools are available to assay human genetic variation contributing to inherited disease syndromes. A method is disclosed for prioritization of genetic variants, and identification of disease genes, using network modeling of gene associations.