Positive-Unlabelled Learning for Identifying New Candidate Dietary Restriction-related Genes among Ageing-related Genes

2024-06-14

Comput. Biol. Medicine

AI Summary

Identifying diet restriction (DR)-associated genes implicated in aging remains challenging, particularly due to the absence of reliable negative labels. Method: This study pioneers the application of positive-unlabeled (PU) learning to DR–aging gene association mining, circumventing reliance on explicitly labeled negatives. We develop an end-to-end predictive framework integrating multi-source functional features, pathway enrichment statistics, and ensemble classifiers—specifically random forest and gradient boosting. Contribution/Results: The PU-learning approach significantly enhances model robustness and generalizability over conventional supervised methods. Experimental validation yields 27 high-confidence novel DR-associated candidate genes, eight of which are independently corroborated by published experimental studies. The model achieves an AUC of 0.89, outperforming state-of-the-art baseline methods. This work establishes a scalable computational paradigm for elucidating DR intervention mechanisms and delivers a high-quality, biologically validated resource of candidate genes for aging research.

Technology Category

Constraint Satisfaction and Optimization: Constraint Learning and AcquisitionMachine Learning: Active LearningData Mining & Knowledge Management: Intelligent Query Processing

Application Category

Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for rankingGraph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphsWeb Mining and Content Analysis: Robustness and generalizability of Web mining methods

Problem

• Identifies new DR-related genes among ageing-related genes.
• Improves reliability of negative examples in gene classification.
• Reduces computational cost and enhances predictive accuracy.

Innovation

• Positive-Unlabelled Learning for gene prioritisation
• KNN-inspired approach selects reliable negative examples
• Classifier differentiates DR-related and non-DR-related genes