FSBLUP: a novel strategy of fusion similarity matrix construction via optimally integrating intermediate omics data to enhance genomic prediction
Abstract
The increasing availability of multi-omics data is promising in enhancing genomic prediction in breeding and human genetics. However, integrating multi-omics data into genomic prediction models remains challenging due to complex relationships between omics layers and phenotypic outcomes. We propose Fusion Similarity Best Linear Unbiased Prediction (FSBLUP), a novel strategy that integrates genomic and intermediate omics data using a unified similarity matrix approach. FSBLUP systematically estimates how different omics layers contribute to phenotypic variation via machine-learning-optimized parameters that capture underlying genetic architecture of complex traits. FSBLUP demonstrates greater predictive accuracy than existing methods, as validated through theoretical and practical evaluations.
Fields of Study & Subjects
Genetic and phenotypic traits in livestock
Genetic Mapping and Diversity in Plants and Animals
Genetic Associations and Epidemiology
Publication details
- Journal / source
- Genome biology
- Publisher
- Springer Science and Business Media LLC
- Volume
- 27
- Issue
- 1
- Pages
- 27–27
- Publication date
- 14 January 2026
- Language
- EN