A comparison of graph- and kernel-based -omics data integration algorithms for classifying complex traits.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-throughput sequencing generates vast multi-omics data for complex disease studies.
• Existing algorithms often analyze single data sources, limiting holistic health insights.
• Integrating multiple data sources is crucial for a comprehensive understanding of human health and diseases.

Purpose of the Study:

• To comprehensively compare the performance of data integration algorithms for binary trait classification.
• To evaluate graph-based and kernel-based integration approaches using real-world disease datasets.

Main Methods:

• Focused on two classes: graph-based and kernel-based integration algorithms.
• Compared seven specific algorithms: graph-based semi-supervised learning, graph sharpening integration, composite association network, Bayesian network, semi-definite programming-support vector machine (SDP-SVM), relevance vector machine (RVM), and Ada-boost relevance vector machine.
• Evaluated algorithms using hypertension and two cancer datasets, assessing classification accuracy and computation time.

Main Results:

• Kernel-based algorithms generally yield higher classification accuracy but demand longer computation times.
• Graph-based algorithms offer faster computation but may result in less complex models.
• Specific algorithms like composite association network, RVM, and Ada-boost RVM demonstrated superior performance.

Conclusions:

• Composite association network, Relevance Vector Machine (RVM), and Ada-boost RVM are recommended for multi-omics data integration.
• Provides guidance on selecting appropriate algorithms for integrating diverse biological data sources for improved disease classification.