Reference

Everything the package exports, built from the docstrings.

The pipeline

pace_fit runs all four stages and returns one object. The stages each take a fit and return it with one more layer attached, for when you want to stop partway or re-run a later one with different settings.

pace.pace_fit	Fit PACE and attach every reporting layer.
pace.pace_model	Fit the PACE model, without the reporting layers.
pace.pace_shrink	Shrink the neighbour slopes and attach them to the fit.
pace.pace_decompose	Decompose the variance and attach both tables to the fit.
pace.pace_drivers	Rank each pair's drivers and attach the tables to the fit.

Readouts

pace.neighbour_slopes	The shrunken slopes, with a local false sign rate per effect.
pace.variance_decomposition	The per-gene variance decomposition.
pace.top_drivers	The per-pair driver tables, or one pair's.
pace.pair_variance	Each neighbour's share of a focal type's spatial variance.
pace.cell_contamination	Per cell: the contamination loading, and the fraction it explains.
pace.anchor_genes	The genes that most mark each cell type, which anchor its contamination.
pace.cell_types	The cell types, in the order they index the random-effect rows.
pace.pace_params	What the fit was run with.
pace.fit_state	The solver's own output: coefficients, errors, dispersions, loadings.
pace.fit_statistics	The per-cell-type statistics, rebuilding the means a gene block at a time.
pace.fitted_means	Rebuild the fitted means for a block of genes.

Figures

pace.plot_decomposition	Stacked bars of each focal type's variance blocks, as R draws them.
pace.plot_pair_heatmap	The pair heatmap, oriented as R draws it: FOCAL on x, NEIGHBOUR on y.
pace.plot_drivers	The top genes mediating one pair, as R draws them.
pace.plot_proximity	A gene's raw counts against how many of a neighbour type a cell has.
pace.plot_shrinkage	What the shrinkage did: the raw slope against the shrunken one.

Building and inspecting a fit

pace.build_neighbourhood	The kernel-weighted neighbour abundances, before any fitting.
pace.ambient_field	The sparse cross-cell-type contamination field, in compressed-column form.
pace.stash_pace	Copy a fit's readouts into adata.uns, leaving the solver state behind.
pace.compact_fit	Drop what a saved fit does not need, keeping every readout.

Objects

pace.PaceFit	A fitted model, with the pieces the readouts need.
pace.FitStatistics	Per-(cell type, gene) summaries of the fitted means.

The numerical core

Single stages, for anyone who wants one on its own rather than a whole fit.

pace.mash	Canonical multivariate adaptive shrinkage for one slice.
pace.driver_scores	Driver scores for one focal-neighbour pair.
pace.pair_variance_pratt	Pratt attribution of spatial variance across a focal type's neighbours.