Package 'SpatialCPie'

SpatialCPie: Cluster analysis of Spatial Transcriptomics data

Description

SpatialCPie is an R package designed to facilitate cluster evaluation for spatial transcriptomics data by providing intuitive visualizations that display the relationships between clusters in order to guide the user during cluster identification and other downstream applications. The package is built around a shiny "gadget" to allow the exploration of the data with multiple plots in parallel and an interactive UI. The user can easily toggle between different cluster resolutions in order to choose the most appropriate visual cues.

Author(s)

Maintainer: Joseph Bergenstraahle [email protected]

---

Parse spot detector output

Description

Parses the output from the ST spot detector tool for use with SpatialCPie.

Usage

parseSpotFile(file)

Arguments

file

spot file

Value

data.frame with columns "x" and "y" specifying the pixel coordinates of each spot

Examples

## Create spot file
data <- rbind(
    c(7, 18, 7.00, 18.07, 563.2, 947.0),
    c(8, 11, 8.00, 11.04, 612.5, 627.7)
)
filename <- tempfile()
write.table(
    data,
    file = filename,
    sep = "\t",
    quote = FALSE,
    col.names = c("x", "y", "new_x", "new_y", "pixel_x", "pixel_y")
)

## Parse spot file
parseSpotFile(filename)

## Delete spot file
unlink(filename)

---

Run SpatialCPie

Description

Runs the SpatialCPie gadget.

Usage

runCPie(counts, image = NULL, spotCoordinates = NULL,
  margin = "spot", resolutions = 2:4,
  assignmentFunction = function(k, x) kmeans(x, centers = k)$cluster,
  view = NULL)

Arguments

counts	gene count matrix or a SummarizedExperiment-class object containing count values.
image	image to be used as background to the plot.
spotCoordinates	data.frame with pixel coordinates. The rows should correspond to the columns (spatial areas) in the count file.
margin	which margin to cluster.
resolutions	numeric vector specifying the clustering resolutions.
assignmentFunction	function to compute cluster assignments.
view	viewer object.

Value

a list with the following items:

• "clusters": Cluster assignments (may differ from assignments)

• "clusterGraph": The cluster tree ggplot object

• "arrayPlot": The pie plot ggplot objects

Examples

if (interactive()) {
    options(device.ask.default = FALSE)

    ## Set up coordinate system
    coordinates <- as.matrix(expand.grid(1:10, 1:10))

    ## Generate data set with three distinct genes generated by three
    ## distinct cell types
    profiles <- diag(rep(1, 3)) + runif(9)
    centers <- cbind(c(5, 2), c(2, 8), c(8, 2))
    mixes <- apply(coordinates, 1, function(x) {
        x <- exp(-colSums((centers - x) ^ 2) / 50)
        x / sum(x)
    })
    means <- 100 * profiles %*% mixes
    counts <- matrix(rpois(prod(dim(means)), means), nrow = nrow(profiles))
    colnames(counts) <- apply(
        coordinates,
        1,
        function(x) do.call(paste, c(as.list(x), list(sep = "x")))
    )
    rownames(counts) <- paste("gene", 1:nrow(counts))

    ## Run SpatialCPie
    runCPie(counts)
}

---