Gene Co-Expression Network: Soft-Thresholding for Scale-Free Topology (2D)

Real gene expression data is noisy: a raw correlation matrix between thousands of genes is nearly a fully-connected "hairball", with no usable structure. WGCNA (Weighted Gene Co-Expression Network Analysis) fixes this with a single trick — raise every correlation to a power β before treating it as an edge weight. This 2D simulator builds a small synthetic expression dataset with four hidden co-expression modules, computes the real Pearson correlation network between genes, and lets you drag that soft-thresholding power live: low β keeps everything densely connected, higher β prunes weak edges exponentially faster than strong ones until a handful of hub genes emerge and a live log-log power-law fit (R²) confirms the network has become scale-free — the same diagnostic real WGCNA pipelines use to choose β on actual RNA-seq data. The network is laid out with a live 2D force simulation you can pan and zoom to inspect module structure up close.