BiomiX

BiomiX enables integrated analysis of multi-omics data from two cohorts, incorporating transcriptomics, metabolomics, and methylomics with subsequent Multi-Omics Factor Analysis to identify shared sources of variation.

Key Features:

• Transcriptomics analysis: Uses DESeq2 and Limma packages for differential expression analysis.
• Metabolomics analysis: Quantifies peak differences via the Wilcoxon test with False Discovery Rate correction, with Liquid Chromatography-Mass Spectrometry untargeted metabolomics annotation supported by mass-to-charge ratio in the CEU Mass Mediator database and fragmentation spectra in the TidyMass package.
• Methylomics analysis: Performed using the ChAMP R package.
• Multi-Omics Factor Analysis integration: Identifies shared sources of variation across omics data, with fine-tuned model optimization for factor number selection and tools to interpret discriminative factors.
• Biological process exploration: Integrates EnrichR and GSEA for pathway analysis and subgroup analysis based on user-defined gene panels.
• Literature-based factor interpretation: Performs bibliography research on Pubmed to identify articles most related to discriminant factor contributors.
• Clinical correlation: Correlates discriminant MOFA factors with clinical data and explores top contributing pathways.

Scientific Applications:

• Multi-omics data integration: Enables integrated analysis of transcriptomics, metabolomics, and methylomics data from two cohorts.
• Condition subtyping: Enhances biological condition classification through subgroup analysis based on user-defined gene panels.
• Factor interpretation: Guides interpretation of discriminative MOFA factors through literature mining and clinical correlation.

Methodology:

BiomiX uses DESeq2/Limma for transcriptomics, Wilcoxon test with FDR correction for metabolomics peak differences, CEU Mass Mediator database and TidyMass package for metabolomics annotation, ChAMP R package for methylomics, and Multi-Omics Factor Analysis for cross-omics integration with fine-tuned factor number optimization.