#36 · Single-cell

Deconvolving cell-state structure from single-cell, chromatin and spatial data

The shared thread across these preprints is recovering structured cell-state information when an external axis, such as cell-cycle phase, ordered experimental conditions, or tissue location, organizes the measurements. CyCLOPS deconvolves genome-wide chromatin occupancy across the cell cycle from MNase-seq of synchronized yeast. SPARC reconstructs directional single-cell trajectories across ordered conditions using a graph-based shortest-path formulation constrained by biological ordering. The osteosarcoma study combines single-cell and spatial transcriptomics with mtDNA lineage tracing, where a WF-HMT model places COL3A1+ cells as the ancestral state. What is new is the explicit use of temporal or spatial ordering as a constraint in inference, rather than treating states as independent. The osteosarcoma item addresses lineage and microenvironment questions more than program decomposition itself.