PRECAST: DLPFC Single Sample Analysis

This vignette introduces the PRECAST workflow for the analysis of single spatial transcriptomics dataset. The workflow consists of three steps

Independent preprocessing and model setting
Probabilistic embedding and clustering using PRECAST model
Downstream analysis (i.e. visualization of clusters and embeddings)

We demonstrate the use of PRECAST to one human dorsolateral prefrontal cortex Visium data that are here, which can be downloaded to the current working path by the following command:

githubURL <- "https://github.com/feiyoung/PRECAST/blob/main/vignettes_data/dlpfc_151672.rda?raw=true"
download.file(githubURL, "dlpfc_151672.rda", mode = "wb")

Then load to R

load("dlpfc_151672.rda")

The package can be loaded with the command:

library(PRECAST)
library(Seurat)

Compare PRECAST with DR-SC in analyzing one sample

First, we view the the spatial transcriptomics data with Visium platform.

dlpfc_151672  ## a list including two Seurat object

check the meta data that must include the spatial coordinates named “row” and “col”, respectively. If the names are not, they are required to rename them.

meta_data <- dlpfc_151672@meta.data
all(c("row", "col") %in% colnames(meta_data))  ## the names are correct!
head(meta_data[, c("row", "col")])

Prepare the PRECASTObject.

Create a PRECASTObj object to prepare for PRECAST models. Here, we only show the HVGs method to select the 2000 highly variable genes, but users are able to choose more genes or also use the SPARK-X to choose the spatially variable genes.

set.seed(2023)
library(PRECAST)
preobj <- CreatePRECASTObject(seuList = list(dlpfc_151672), selectGenesMethod = "HVGs", gene.number = 2000)  #

Add the model setting

## check the number of genes/features after filtering step
preobj@seulist
## Add adjacency matrix list for a PRECASTObj object to prepare for PRECAST model fitting.
PRECASTObj <- AddAdjList(preobj, platform = "Visium")
## Add a model setting in advance for a PRECASTObj object. verbose =TRUE helps outputing the
## information in the algorithm.
PRECASTObj <- AddParSetting(PRECASTObj, Sigma_equal = FALSE, coreNum = 1, maxIter = 30, verbose = TRUE)

Fit PRECAST

For function PRECAST, users can specify the number of clusters K or set K to be an integer vector by using modified BIC(MBIC) to determine K. Here, we use user-specified number of clusters.

### Given K
PRECASTObj <- PRECAST(PRECASTObj, K = 7)

Use the function SelectModel() to re-organize the fitted results in PRECASTObj.

## backup the fitting results in resList
resList <- PRECASTObj@resList
PRECASTObj <- SelectModel(PRECASTObj)
ari_precast <- mclust::adjustedRandIndex(PRECASTObj@resList$cluster[[1]], PRECASTObj@seulist[[1]]$layer_guess_reordered)

Other options

Users are also able to set multiple K, then choose the best one

PRECASTObj2 <- AddParSetting(PRECASTObj, Sigma_equal = FALSE, coreNum = 4, maxIter = 30, verbose = TRUE)  # set 4 cores to run in parallel.
PRECASTObj2 <- PRECAST(PRECASTObj2, K = 5:8)
## backup the fitting results in resList
resList2 <- PRECASTObj2@resList
PRECASTObj2 <- SelectModel(PRECASTObj2)
str(PRECASTObj2@resList)
mclust::adjustedRandIndex(PRECASTObj2@resList$cluster[[1]], PRECASTObj2@seulist[[1]]$layer_guess_reordered)

Besides, user can also use different initialization method by setting int.model, for example, set int.model=NULL; see the functions AddParSetting() and model_set() for more details.

Put the reults into a Seurat object seuInt.

seuInt <- PRECASTObj@seulist[[1]]
seuInt@meta.data$cluster <- factor(unlist(PRECASTObj@resList$cluster))
seuInt@meta.data$batch <- 1
seuInt <- Add_embed(PRECASTObj@resList$hZ[[1]], seuInt, embed_name = "PRECAST")
posList <- lapply(PRECASTObj@seulist, function(x) cbind(x$row, x$col))
seuInt <- Add_embed(posList[[1]], seuInt, embed_name = "position")
Idents(seuInt) <- factor(seuInt@meta.data$cluster)

seuInt
## The low-dimensional embeddings obtained by PRECAST are saved in PRECAST reduction slot.

Save the spatial and tSNE scatter plots for clusters from PRECAST

p_sp1 <- SpaPlot(seuInt, item = "cluster", point_size = 3, combine = F)[[1]] + cowplot::theme_cowplot() +
    ggplot2::ggtitle(paste0("PRECAST: ARI=", round(ari_precast, 2))) + ggplot2::xlab("row") + ggplot2::ylab("col")
seuInt <- AddTSNE(seuInt, n_comp = 2)
p_tsne <- dimPlot(seuInt, item = "cluster")
p_tsne <- p_tsne + cowplot::theme_cowplot() + ggplot2::ggtitle("PRECAST")

Fit DR-SC and Plot the spatial and tSNE scatter plots for clusters

seu_drsc <- DR.SC::DR.SC(PRECASTObj@seulist[[1]], K = 7, verbose = T)
ari_drsc <- mclust::adjustedRandIndex(seu_drsc$spatial.drsc.cluster, PRECASTObj@seulist[[1]]$layer_guess_reordered)
p_tsne_drsc <- DR.SC::drscPlot(seu_drsc)
p_tsne_drsc <- p_tsne_drsc + ggplot2::ggtitle("DR-SC")
p_sp2 <- DR.SC::spatialPlotClusters(seu_drsc) + cowplot::theme_cowplot() + ggplot2::ggtitle(paste0("DR-SC ARI=",
    round(ari_drsc, 2)))

Compare the clustering performance of PRECAST and DR-SC.

drawFigs(list(p_sp1, p_sp2), layout.dim = c(1, 2))

Compare the tSNE visualiztion performance of PRECAST and DR-SC.

drawFigs(list(p_tsne, p_tsne_drsc), layout.dim = c(1, 2))

Session Info

sessionInfo()
#> R version 4.4.2 (2024-10-31)
#> Platform: x86_64-pc-linux-gnu
#> Running under: Ubuntu 24.04.1 LTS
#> 
#> Matrix products: default
#> BLAS:   /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3 
#> LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/libopenblasp-r0.3.26.so;  LAPACK version 3.12.0
#> 
#> locale:
#>  [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
#>  [3] LC_TIME=en_US.UTF-8        LC_COLLATE=C              
#>  [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
#>  [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
#>  [9] LC_ADDRESS=C               LC_TELEPHONE=C            
#> [11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       
#> 
#> time zone: Etc/UTC
#> tzcode source: system (glibc)
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> other attached packages:
#> [1] rmarkdown_2.29
#> 
#> loaded via a namespace (and not attached):
#>  [1] digest_0.6.37     R6_2.5.1          fastmap_1.2.0     xfun_0.49        
#>  [5] maketools_1.3.1   cachem_1.1.0      knitr_1.49        htmltools_0.5.8.1
#>  [9] buildtools_1.0.0  lifecycle_1.0.4   cli_3.6.3         sass_0.4.9       
#> [13] jquerylib_0.1.4   compiler_4.4.2    sys_3.4.3         tools_4.4.2      
#> [17] evaluate_1.0.1    bslib_0.8.0       yaml_2.3.10       formatR_1.14     
#> [21] jsonlite_1.8.9    rlang_1.1.4

- Compare PRECAST with DR-SC in analyzing one sample