Bayesian spatial finite mixture models for identification of cell sub-populations in multi-sample spatial transcriptomics experiments
maple is a Bayesian statistical framework for modeling multi-sample spatial transcriptomics data sets. maple is designed to be used within standard Seurat workflows, and the user may specify to use principal components (PCs), highly variable genes (HVGs), spatially variable genes (SVGs), or custom cell/cell-spot embeddings such as those generated by RESEPT. A pre-print publication of maple can be found on bioRxiv.
You may install the stable package version from CRAN, or the development version as follows:
devtools::install_github("carter-allen/maple")
Installation time is under 10 minutes for all depedent R packages. maple has been tested on Windows, macOS, and Linux systems as detailed here
- Multi-Sample Anterior Mouse Brain
- Multi-Sample Posterior Mouse Brain
- Multi-Sample Anterior & Posterior Mouse Brain
Below are command lines for analysis of a 2-sample anterior mouse brain data set sequenced with the 10X Visium platform. Each model takes < 10 minutes to run for 2,000 iterations on a M1 iMac.
# Single cell code
library(Seurat)
library(SeuratData)
library(maple)
# Load data
InstallData("stxBrain")
brain1 <- LoadData("stxBrain", type = "anterior1")
brain2 <- LoadData("stxBrain", type = "anterior2")
brain1 <- SCTransform(brain1, assay = "Spatial", verbose = FALSE)
brain2 <- SCTransform(brain2, assay = "Spatial", verbose = FALSE)
brain <- merge(brain1,brain2)
DefaultAssay(brain) <- "SCT"
VariableFeatures(brain) <- c(VariableFeatures(brain1),VariableFeatures(brain2))
brain <- RunPCA(brain)
save(brain, file = paste0(data_dir,"brain_anterior12_merged.RData"))
load(paste0(data_dir,"brain_anterior12_merged.RData"))
# fit maple model
# PC features
brain_fit_PCs <- fit_maple(brain,K = 6,emb = "PCs")
# plot results
brain$maple_labels_PCs_K6 <- as.factor(brain_fit_PCs$z)
Idents(brain) <- "maple_labels_PCs_K6"
SpatialDimPlot(brain,label = TRUE)
# HVG features
brain_fit_HVGs <- fit_maple(brain,K = 6,emb = "HVGs")
# plot results
brain$maple_labels_HVGs_K6 <- as.factor(brain_fit_HVGs$z)
Idents(brain) <- "maple_labels_HVGs_K6"
SpatialDimPlot(brain,label = TRUE)
# visualize coefficients
spruce::plot_deltas(brain_fit_PCs)
# get scores
post_scores <- spruce::get_scores(brain_fit_PCs)
[email protected] <- cbind([email protected],post_scores)
# plot scores
SpatialFeaturePlot(brain,features = "u_score")
SpatialFeaturePlot(brain,features = "k1_score")
Allen, C., Chang, Y., Ma, Q., & Chung, D. (2022). MAPLE: A Hybrid Framework for Multi-Sample Spatial Transcriptomics Data. bioRxiv.