New here? Click through the pipeline below β or hit Play and let it walk you through every stage, from raw single-cell data to the matrices that come out the other end. Matrices small enough to render live (C, Δ, Ξ²) are real interactive charts β hover any cell or dot for the actual type/gene names. Every node also explains how to read its heatmap: what's on each axis, and what the pattern is telling you.
What the pipeline starts with β nothing here has a cell-identity answer yet.
build_all_matrices.py turns the raw inputs into the matrices the solver needs.
cm_visual/solver.py β the alternating optimisation loop, the heart of ConnectionMiner.
What comes out the other end, and gets handed to a biologist.