Enhance spot resolution

Enhanced clustering of a spatial expression dataset to subspot resolution.

spatialEnhance(
  sce,
  q,
  platform = c("Visium", "ST"),
  use.dimred = "PCA",
  d = 15,
  init = NULL,
  init.method = c("spatialCluster", "mclust", "kmeans"),
  model = c("t", "normal"),
  nrep = 2e+05,
  gamma = NULL,
  mu0 = NULL,
  lambda0 = NULL,
  alpha = 1,
  beta = 0.01,
  save.chain = FALSE,
  chain.fname = NULL,
  burn.in = 10000,
  jitter_scale = 5,
  jitter_prior = 0.3,
  verbose = FALSE
)

Arguments

sce	A SingleCellExperiment object containing the spatial data.
q	The number of clusters.
platform	Spatial transcriptomic platform. Specify 'Visium' for hex lattice geometry or 'ST' for square lattice geometry. Specifying this parameter is optional when analyzing SingleCellExperiments processed using readVisium, spatialPreprocess, or spatialCluster, as this information is included in their metadata.
use.dimred	Name of a reduced dimensionality result in reducedDims(sce). If provided, cluster on these features directly.
d	Number of top principal components to use when clustering.
init	Initial cluster assignments for spots.
init.method	If init is not provided, cluster the top d PCs with this method to obtain initial cluster assignments.
model	Error model. ('normal' or 't')
nrep	The number of MCMC iterations.
gamma	Smoothing parameter. (Values in range of 1-3 seem to work well.)
mu0	Prior mean hyperparameter for mu. If not provided, mu0 is set to the mean of PCs over all spots.
lambda0	Prior precision hyperparam for mu. If not provided, lambda0 is set to a diagonal matrix \(0.01 I\).
alpha	Hyperparameter for Wishart distributed precision lambda.
beta	Hyperparameter for Wishart distributed precision lambda.
save.chain	If true, save the MCMC chain to an HDF5 file.
chain.fname	File path for saved chain. Tempfile used if not provided.
burn.in	Number of iterations to exclude as burn-in period. The MCMC iterations are currently thinned to every 100; accordingly burn.in is rounded down to the nearest multiple of 100.
jitter_scale	Controls the amount of jittering. Small amounts of jittering are more likely to be accepted but result in exploring the space more slowly. We suggest tuning jitter_scale so that Ychange is on average around 25%-40%.
jitter_prior	Scale factor for the prior variance, parameterized as the proportion (default = 0.3) of the mean variance of the PCs. We suggest making jitter_prior smaller if the jittered values are not expected to vary much from the overall mean of the spot.
verbose	Log progress to stderr.

Value

Returns a new SingleCellExperiment object. By default, the assays of this object are empty, and the enhanced resolution PCs are stored as a reduced dimensionality result accessible with reducedDim(sce, 'PCA').

Details

The enhanced SingleCellExperiment has most of the properties of the input SCE - rowData, colData, reducedDims - but does not include expression data in counts or logcounts. To impute enhanced expression vectors, please use [enhanceFeatures()] after running spatialEnhance.

The colData of the enhanced SingleCellExperiment includes the following columns to permit referencing the subspots in spatial context and linking back to the original spots:

• spot.idx: Index of the spot this subspot belongs to (with respect to the input SCE).

• subspot.idx: Index of the subspot within its parent spot.

• spot.row: Array row of the subspot's parent spot.

• spot.col: Array col of the subspot's parent spot.

• row: Array row of the subspot. This is the parent spot's row plus an offset based on the subspot's position within the spot.

• col: Array col of the subspot. This is the parent spot's col plus an offset based on the subspot's position within the spot.

• imagerow: Pixel row of the subspot. This is the parent spot's row plus an offset based on the subspot's position within the spot.

• imagecol: Pixel col of the subspot. This is the parent spot's col plus an offset based on the subspot's position within the spot.

See also

spatialCluster for clustering at the spot level before enhancing, clusterPlot for visualizing the cluster assignments, enhanceFeatures for imputing enhanced expression, and mcmcChain for examining the full MCMC chain associated with the enhanced clustering. .

Examples

set.seed(149)
sce <- exampleSCE()
sce <- spatialCluster(sce, 7, nrep=100, burn.in=10)
#> Neighbors were identified for 96 out of 96 spots.
#> Fitting model...
#> Calculating labels using iterations 10 through 100.
enhanced <- spatialEnhance(sce, 7, nrep=100, burn.in=10)
#> Calculating labels using iterations 0 through 100.