model_search: Compare model fits with different models

In sales-lab/spatialDE: R wrapper for SpatialDE

Compare model fits with different models

Description

Classify DE genes to interpretable fitting classes.

Usage

model_search(x, coordinates, de_results, qval_thresh = 0.05, verbose = FALSE)

Arguments

x	matrix-like object of normalized counts. E.g. resulting from regress_out().
coordinates	data.frame with sample coordinates. Each row is a sample, the columns with coordinates must be named 'x' and 'y'.
de_results	data.frame resulting from run().
qval_thresh	numeric scalar, specifying the q-value significance threshold to filter de_results. Only rows in de_results with qval < qval_thresh will be kept. To disable, set qval_thresh = NULL.
verbose	logical controlling the display of the progress bar.

Value

data.frame of model_search results.

References

Svensson, V., Teichmann, S. & Stegle, O. SpatialDE: identification of spatially variable genes. Nat Methods 15, 343–346 (2018). https://doi.org/10.1038/nmeth.4636

Examples

## Mock up a SpatialExperiment object wit 400 cells and 3 genes
set.seed(42)
mock <- mockSVG(size = 20, tot_genes = 3, de_genes = 1)

stabilized <- stabilize(mock$counts)
sample_info <- mock$coordinates
sample_info$total_counts <- colSums(mock$counts)
regressed <- regress_out(counts = stabilized, sample_info = sample_info)

## Run SpatialDE
de_results <- run(regressed, coordinates = mock$coordinates)

## Run model search
ms_results <- model_search(
    x = regressed,
    coordinates = mock$coordinates,
    de_results = de_results,
    qval_thresh = NULL
)

sales-lab/spatialDE documentation built on Feb. 12, 2024, 2:47 p.m.