Combining Partially Overlapping Multi-Omics Data in Databases Using Relationship Matrices.

Area of Science:

• Genomics
• Quantitative Genetics
• Bioinformatics

Background:

• Genomics technologies generate vast amounts of sequence and phenotype data.
• Managing and analyzing complex, high-dimensional omics data presents significant challenges.
• Integrating diverse datasets is essential for advancing genetic research and trait discovery.

Purpose of the Study:

• To address the challenge of combining high-dimensional and unbalanced omics data.
• To propose a novel covariance-based method for integrating partially overlapping datasets in the genotype-to-phenotype spectrum.
• To demonstrate the utility of data harmonization for genomic prediction and understanding trait relationships.

Main Methods:

• Development of a covariance-based method for combining partial datasets.
• Application of the method to harmonize relationship/covariance matrices.
• Utilizing the harmonized data for genomic prediction with heterogeneous marker data.
• Combining data from multiple phenotypic experiments.

Main Results:

• The proposed method is advantageous compared to feature imputation approaches.
• Demonstrated successful genomic prediction using harmonized, heterogeneous marker data.
• Enabled inferences about previously unobserved trait relationships by combining multiple phenotypic experiments.
• Showcased the possibility of harmonizing datasets to enhance information across data resources.

Conclusions:

• Data harmonization is a viable strategy to improve information across gene banks and data repositories.
• The covariance-based method effectively combines omics data for enhanced genomic insights.
• This approach accelerates the identification of key traits and improves the understanding of quantitative genetics.