🧫 2D Spatial Transcriptomics Spot Deconvolution: EM Estimation of Cell-Type Proportions from Mixed Gene Counts with the Cramer-Rao Limit
True and estimated cell-type pies on a 12 x 12 spot grid with an error map, live EM iterations, likelihood and RMSE plots, spatial smoothing option and six independent checks (EM monotone, exact noise-free recovery, Cramer-Rao bound, Poisson statistics, 1/sqrt(depth) error).
⚙ Under the hood
True and estimated cell-type pies on a 12 x 12 spot grid with an error map, live EM iterations, likelihood and RMSE plots, spatial smoothing option and six independent checks (EM monotone, exact noise-free recovery, Cramer-Rao bound, Poisson statistics, 1/sqrt(depth) error).
Spatial TranscriptomicsDeconvolutionEM AlgorithmCell TypesCramer-Rao BoundPoisson CountsVisiumBioinformatics
2D · HTML5 Canvas 2D · 60 FPS target · runs fully client-side, no install