Phenotyping Graph Model for Gene Ranking From Heterogeneous Symptoms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current phenotyping practices in precision medicine suffer from inconsistent and heterogeneous descriptions of symptoms, leading to challenges in accurately identifying genes associated with observed symptoms, which hinders the full exploitation of clinical data.
Innovation Solution
A device and method that utilize a gene-phenotype matrix to calculate the probability of gene-symptom associations, incorporating correction vectors and normalization weights, and employ graph-based models to standardize phenotyping and identify the most likely genes associated with observed symptoms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If physicians use predefined ontologies such as HPO to describe symptoms, then standardization of clinical descriptions is improved, but heterogeneity in symptom description remains due to varying physician interpretations
Solution Approach 1:
The patent introduces an intermediary computational model that mediates between physician descriptions and gene associations. The model processes heterogeneous symptom descriptions through a standardized representation layer, enabling consistent gene prioritization despite varying input formats and interpretations.
Solution Approach 2:
The patent transforms symptom descriptions into quantitative parameters through probability scores derived from the gene-phenotype matrix. By converting qualitative descriptions into standardized probability values, the system achieves consistent comparison and ranking of gene associations across different symptom reports.
2Quantity of substance
If current algorithms use ontology structure to extract symptom-gene associations, then available data is utilized, but computational complexity increases and speed is reduced
Solution Approach 1:
The patent pre-computes and stores the gene-phenotype matrix with probability scores during system initialization. This preliminary action allows the system to directly query pre-processed association strengths during actual gene prioritization, avoiding complex real-time computations and significantly improving processing speed.
Solution Approach 2:
The patent replaces complex algorithmic processing of ontology structures with direct matrix-based probability calculations. By substituting intricate computational graph traversals with straightforward matrix operations, the system maintains comprehensive data utilization while dramatically reducing computational complexity and increasing speed.
3Measurement precision
If symptom-gene associations are ranked based on multiple factors including correction vectors, then accuracy of gene identification is improved, but device complexity increases
Solution Approach 1:
The patent segments the gene prioritization process into distinct modular components: the gene-phenotype matrix for base associations, correction vectors for specific adjustments, and normalization weights for scaling. This segmentation allows each component to be independently computed, stored, and applied, reducing overall system complexity while maintaining high accuracy through systematic combination of factors.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present invention relates to a computer-implemented method for decision support of a user to standardize phenotyping in genomic analysis of a subject, wherein the method comprises: - receiving a list of symptoms comprising at least one symptom observed for the subject; - receiving a first graph comprising nodes and weighted links; - receiving a second graph being previously obtained applying a matrix factorization to a gene-symptom matrix; - outputting at least one gene associated to the list of symptoms based on the first graph and on the second graph.