Run Canek on a toy example

library(Canek)

# Functions
## Function to plot the pca coordinates
plotPCA <- function(pcaData = NULL, label = NULL, legPosition = "topleft"){
  col <- as.integer(label) 
  plot(x = pcaData[,"PC1"], y = pcaData[,"PC2"],
       col = as.integer(label), cex = 0.75, pch = 19,
       xlab = "PC1", ylab = "PC2")
  legend(legPosition,  pch = 19,
         legend = levels(label), 
         col =  unique(as.integer(label)))
}

Load the data

On this toy example we use the two simulated batches included in the SimBatches data from Canek’s package. SimBatches is a list containing:

batches: Simulated scRNA-seq datasets with genes (rows) and cells (columns). Simulations were performed using Splatter.
cell_type: a factor containing the celltype labels of the batches

lsData <- list(B1 = SimBatches$batches[[1]], B2 = SimBatches$batches[[2]])
batch <- factor(c(rep("Batch-1", ncol(lsData[[1]])),
                  rep("Batch-2", ncol(lsData[[2]]))))
celltype <- SimBatches$cell_types
table(batch)
#> batch
#> Batch-1 Batch-2 
#>     631     948
table(celltype)
#> celltype
#> Cell Type 1 Cell Type 2 Cell Type 3 Cell Type 4 
#>        1451          53          38          37

PCA before correction

We perform the Principal Component Analysis (PCA) of the joined datasets and scatter plot the first two PCs. The batch-effect causes cells to group by batch.

data <- Reduce(cbind, lsData)
pcaData <- prcomp(t(data), center = TRUE, scale. = TRUE)$x

plotPCA(pcaData = pcaData, label = batch, legPosition = "bottomleft")

plotPCA(pcaData = pcaData, label = celltype, legPosition = "bottomleft")

Run Canek

We correct the toy batches using the function RunCanek. This function accepts:

List of matrices
Seurat object
List of Seurat objects
SingleCellExperiment object
List of SingleCellExperiment objects

On this example we use the list of matrices created before.

data <- RunCanek(lsData)

PCA after correction

We perform PCA of the corrected datasets and plot the first two PCs. After correction, the cells group by their corresponding cell type.

pcaData <- prcomp(t(data), center = TRUE, scale. = TRUE)$x

plotPCA(pcaData = pcaData, label = batch, legPosition = "topleft")

plotPCA(pcaData = pcaData, label = celltype, legPosition = "topleft")

Session info

sessionInfo()
#> R version 4.1.3 (2022-03-10)
#> Platform: x86_64-apple-darwin13.4.0 (64-bit)
#> Running under: macOS Big Sur/Monterey 10.16
#> 
#> Matrix products: default
#> BLAS/LAPACK: /Users/martin/miniconda3/envs/R_4.1.3/lib/libopenblasp-r0.3.18.dylib
#> 
#> locale:
#> [1] C/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8
#> 
#> attached base packages:
#> [1] stats     graphics  grDevices utils     datasets  methods   base     
#> 
#> other attached packages:
#> [1] Canek_0.2.5
#> 
#> loaded via a namespace (and not attached):
#>  [1] Rcpp_1.0.10          highr_0.10           DEoptimR_1.0-11     
#>  [4] bslib_0.4.2          compiler_4.1.3       bluster_1.4.0       
#>  [7] jquerylib_0.1.4      class_7.3-21         prabclus_2.3-2      
#> [10] BiocNeighbors_1.12.0 numbers_0.8-5        tools_4.1.3         
#> [13] digest_0.6.31        mclust_6.0.0         jsonlite_1.8.4      
#> [16] evaluate_0.20        lattice_0.20-45      pkgconfig_2.0.3     
#> [19] rlang_1.0.6          Matrix_1.5-3         igraph_1.3.5        
#> [22] cli_3.6.0            rstudioapi_0.14      yaml_2.3.7          
#> [25] parallel_4.1.3       xfun_0.37            fastmap_1.1.0       
#> [28] knitr_1.42           cluster_2.1.4        sass_0.4.5          
#> [31] S4Vectors_0.32.4     fpc_2.2-10           diptest_0.76-0      
#> [34] nnet_7.3-18          stats4_4.1.3         grid_4.1.3          
#> [37] robustbase_0.95-0    R6_2.5.1             flexmix_2.3-18      
#> [40] BiocParallel_1.28.3  rmarkdown_2.20       irlba_2.3.5.1       
#> [43] kernlab_0.9-32       magrittr_2.0.3       matrixStats_0.63.0  
#> [46] modeltools_0.2-23    htmltools_0.5.4      BiocGenerics_0.40.0 
#> [49] MASS_7.3-58.3        cachem_1.0.6         FNN_1.1.3.1