scNPM - Single-Cell Network Propensity Matching

Estimates and compares heterogeneous gene conditional-dependence networks from single-cell expression data. A latent Gaussian mixture and probit observation mechanism represent cellular heterogeneity and zero observations. Protein interaction information supplies edge-specific prior weights, while network propensity matching links cell clusters across conditions. Staged inference fixes correspondence after adaptation and jointly models matched network edges with a four-state spike-and-slab prior for posterior group and directional differential edge selection. Functions assess balance, record criterion diagnostics, and visualize within-condition and differential networks.