ONION: Functional Approach for Integration of Lipidomics and Transcriptomics Data.

Area of Science:

• Bioinformatics
• Systems Biology
• Genomics
• Metabolomics

Background:

• High-throughput data generation outpaces current bioinformatics analytical capabilities.
• Existing multidimensional statistical methods, like canonical correlation analysis (CCA), struggle with integrating diverse data types (e.g., transcriptomics, metabolomics) due to small sample sizes relative to data dimensionality.
• Current modifications to statistical approaches often involve data simulation or variable pruning, leading to loss of information or unreliable statistical measures.

Purpose of the Study:

• To develop and validate a novel bioinformatics approach for integrating multi-omics data, specifically transcriptomics and metabolomics.
• To overcome the limitations of traditional statistical methods in analyzing high-throughput biological data with small sample sizes.
• To improve the detection and interpretation of metabolite-gene associations, particularly in understanding metabolic responses.

Main Methods:

• Utilized verified or putative molecular interactions and functional associations to guide data analysis.
• Implemented a workflow involving data set partitioning, intra-group statistical analysis, and pathway/network analysis for robust association discovery.
• Applied canonical correlation analysis (CCA) and other multivariate models to integrated datasets, leveraging pathway and network information to group variables.

Main Results:

• The novel approach successfully integrated lipidomics and transcriptomics data from public murine nutrigenomics datasets.
• Demonstrated improved detection of genes related to lipid metabolism compared to traditional statistical methods alone, with a higher percentage of explained variance (95% vs. 75-80%).
• Identified new, robust metabolite-gene associations relevant to lipid metabolism, enhancing biological understanding.

Conclusions:

• The proposed bioinformatics workflow effectively integrates multi-omics data by incorporating molecular interaction and pathway information.
• This method overcomes limitations of standard statistical techniques, enabling robust identification of metabolite-gene associations and improved understanding of biological systems.
• The approach offers a significant advancement for leveraging high-throughput data in biological and biomedical research, particularly in nutrigenomics.