OmniHarmonyMULTI-OMICS
Integrated latent-space fusion dashboard — 182 cells across 4 omics layers
5 clusters · 8 factors · 4 layers
LAYER HARMONY?
Transcriptomics
Proteomics
Epigenomics
Metabolomics
Shared Latent Space · UMAP?hover to inspect · click to select
Factor Network?
Immune Acti…Metabolic S…Cell CycleEpigenetic …Signal Casc…Stress Resp…Differentia…Angiogenesis
Factor Loadings
Click a factor node to explore loadings
Cross-Layer Correlations?
⟡
Select a feature from the factor loadings
or click a point on the UMAP to explore
cross-omics correlations
Dataset Summary
Total Cells182
Clusters5
Latent Factors8
Features / Layer20
IntegrationMOFA+
XAI · How Fusion Works

Latent Space Integration — MOFA+ decomposes each omics matrix into shared factors and layer-specific noise. The UMAP is computed from the joint factor matrix.

Factor Loadings — Each factor captures a biological theme (pathway, cell state). High-loading features drive that factor across layers.

Cross-Layer Signals — Correlations use regression weights from the factor model to estimate inter-omics relationships.

OMNIHARMONY · MULTI-OMICS INTEGRATION DASHBOARD · SYNTHETIC DEMO DATA