Skip to contents

Run STdeconvolve reference-free spatial deconvolution

Source: R/RunSTdeconvolve.R
RunSTdeconvolve.Rd

Estimate spot-level topic proportions from a spatial Seurat object using the optional STdeconvolve package.

Usage

RunSTdeconvolve(
  srt,
  assay = NULL,
  layer = "counts",
  features = NULL,
  k = NULL,
  k_candidates = 2:9,
  opt = "min",
  clean_counts = TRUE,
  clean_counts_params = list(),
  restrict_corpus = TRUE,
  restrict_corpus_params = list(),
  fit_lda_params = list(),
  get_beta_theta_params = list(),
  prefix = "STdeconvolve",
  tool_name = "STdeconvolve",
  store_results = TRUE,
  round_counts = TRUE,
  verbose = TRUE,
  ...
)

Arguments

srt

Spatial Seurat object used as the RCTD query.

assay

Assay used in srt. If NULL, the default assay is used.

layer

Assay layer used as STdeconvolve input.

features

Features used for RCTD. If NULL, shared features are used.

k

Number of topics. If NULL, models are fit over k_candidates and STdeconvolve::optimalModel() is used to choose a model.

k_candidates

Candidate topic numbers used when k = NULL.

opt

Optimal model selector passed to STdeconvolve::optimalModel().

clean_counts

Whether to call STdeconvolve::cleanCounts().

clean_counts_params

Additional parameters passed to STdeconvolve::cleanCounts().

restrict_corpus

Whether to call STdeconvolve::restrictCorpus().

restrict_corpus_params

Additional parameters passed to STdeconvolve::restrictCorpus().

fit_lda_params

Additional parameters passed to STdeconvolve::fitLDA().

get_beta_theta_params

Additional parameters passed to STdeconvolve::getBetaTheta().

prefix

Prefix for metadata columns.

tool_name

Name used to store detailed results in srt@tools.

store_results

Whether to store detailed RCTD results in srt@tools.

round_counts

Whether to round non-integer counts before model fitting.

verbose

Whether to print the message. Default is TRUE.

...

Additional parameters passed to the RCTD run step.

Value

A Seurat object with topic proportions in metadata and detailed results stored in srt@tools[[tool_name]] when store_results = TRUE.

Examples

data(visium_human_pancreas_sub)
spatial <- visium_human_pancreas_sub
topic_weights <- data.frame(
  STdeconvolve_prop_topic_1 = seq(0.75, 0.20, length.out = ncol(spatial)),
  STdeconvolve_prop_topic_2 = seq(0.20, 0.70, length.out = ncol(spatial)),
  STdeconvolve_prop_topic_3 = 0.10,
  row.names = colnames(spatial)
)
topic_weights <- topic_weights / rowSums(topic_weights)
spatial <- Seurat::AddMetaData(spatial, topic_weights)
spatial$STdeconvolve_dominant_type <- sub(
  "^STdeconvolve_prop_",
  "",
  colnames(topic_weights)[max.col(topic_weights)]
)
spatial$STdeconvolve_max_prop <- apply(topic_weights, 1, max)

SpatialSpotPlot(
  spatial,
  group.by = "STdeconvolve_dominant_type",
  overlay_image = FALSE,
  coord.cols = c("x", "y")
)

spatial <- RunSTdeconvolve(
  spatial,
  assay = "Spatial",
  features = rownames(spatial)[1:300],
  k = 3,
  verbose = FALSE
)
#> Removing 0 genes present in 100% or more of pixels...
#> 300 genes remaining...
#> Removing 0 genes present in 5% or less of pixels...
#> 300 genes remaining...
#> Restricting to overdispersed genes with alpha = 0.05...
#> Calculating variance fit ...
#> Using gam with k=5...
#> 68 overdispersed genes ... 
#>  Using top 1000 overdispersed genes.
#>  number of top overdispersed genes available: 68
#> Time to fit LDA models was 0.07 mins
#> Computing perplexity for each fitted model...
#> Time to compute perplexities was 0.04 mins
#> Getting predicted cell-types at low proportions...
#> Time to compute cell-types at low proportions was 0 mins
#> Plotting...
#> Warning: Ignoring unknown parameters: `linewidth`
#> Warning: Ignoring unknown parameters: `linewidth`
#> Warning: The `size` argument of `element_line()` is deprecated as of ggplot2 3.4.0.
#> ℹ Please use the `linewidth` argument instead.
#> ℹ The deprecated feature was likely used in the STdeconvolve package.
#>   Please report the issue at
#>   <https://github.com/JEFworks-Lab/STdeconvolve/issues>.
#> `geom_line()`: Each group consists of only one observation.
#> ℹ Do you need to adjust the group aesthetic?
#> `geom_line()`: Each group consists of only one observation.
#> ℹ Do you need to adjust the group aesthetic?

#> Filtering out cell-types in pixels that contribute less than 0.05 of the pixel proportion.
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to max; returning -Inf
#> Warning: no non-missing arguments to max; returning -Inf
STdeconvolvePlot(
  spatial,
  topics = 1:2,
  overlay_image = FALSE,
  coord.cols = c("x", "y")
)

STdeconvolvePlot(
  spatial,
  plot_type = "pie",
  overlay_image = FALSE,
  coord.cols = c("x", "y")
)