Each of the four molecular layers holds 10 measured entities (genes, transcripts, proteins, metabolites), laid out here as four horizontal bands — the same per-layer ellipse coordinates the 3D version places at a fixed camera depth, just read top-to-bottom instead of near-to-far. For the current patient, every entity carries a z-score relative to a healthy reference population. A gene, its transcript, the protein it encodes and a downstream metabolite are joined by a vertical "central dogma" edge; within a band, entities whose simulated correlation exceeds your threshold are also joined by an arc.
The panel computes a weighted multi-omics integration score, the same idea behind similarity-network-fusion approaches used in real precision-medicine pipelines:
R = 100 · Σ wᵢ·mean(|zᵢ|) / Σ wᵢ for i ∈ {genomics, transcriptomics, proteomics, metabolomics}
- Layer weight sliders — set how much each omics layer contributes to the fused risk score wᵢ; a layer at 0% is still shown but excluded from R. The bar panel below the network shows each layer's live contribution wᵢ·mean(|zᵢ|).
- Correlation edge threshold — only within-band entity pairs whose simulated |correlation| exceeds this value draw a connecting arc, showing how tightening the threshold sparsifies each band's sub-network.
- Node size gain — rescales how strongly |z| drives node radius/glow, purely visual, so outliers can be made to pop more or less.
- Node color/size — hue marks the layer (ring colour), fill marks direction/magnitude of z (blue = low, white = normal, red = high), node radius and glow scale with |z|.
- Drag to pan, scroll/pinch to zoom — the network map supports the same free exploration an orbit camera gives the 3D version; "Reset view" recenters it.
- Next Patient — swaps in a different simulated omics profile (healthy baseline, an oncogenic signature skewed toward genomics/proteomics, and a metabolic-syndrome signature skewed toward proteomics/metabolomics) so you can see how the same weights produce very different fused scores depending on where the real biological signal sits.
Real-world relevance: oncology and rare-disease programs increasingly integrate genomic, transcriptomic, proteomic and metabolomic panels per patient rather than reading any single layer alone, because a single-omics test can miss a signal that only shows up as a correlated shift across layers.