Introduction to spatialDE
25 April 2023
1 Introduction
SpatialDE by Svensson et al., 2018, is a method to identify spatially variable genes (SVGs) in spatially resolved transcriptomics data.
2 Installation
You can install spatialDE from Bioconductor with the following code:
if (!requireNamespace("BiocManager", quietly = TRUE)) {
install.packages("BiocManager")
}
BiocManager::install("spatialDE")3 Example: Mouse Olfactory Bulb
Reproducing the example analysis from the original SpatialDE Python package.
library(spatialDE)
library(ggplot2)3.1 Load data
Files originally retrieved from SpatialDE GitHub repository from the following links: https://github.com/Teichlab/SpatialDE/blob/master/Analysis/MouseOB/data/Rep11_MOB_0.csv https://github.com/Teichlab/SpatialDE/blob/master/Analysis/MouseOB/MOB_sample_info.csv
# Expression file used in python SpatialDE.
data("Rep11_MOB_0")
# Sample Info file used in python SpatialDE
data("MOB_sample_info")The Rep11_MOB_0 object contains spatial expression data for
16218 genes on 262 spots, with genes as rows
and spots as columns.
Rep11_MOB_0[1:5, 1:5]
#> 16.92x9.015 16.945x11.075 16.97x10.118 16.939x12.132 16.949x13.055
#> Nrf1 1 0 0 1 0
#> Zbtb5 1 0 1 0 0
#> Ccnl1 1 3 1 1 0
#> Lrrfip1 2 2 0 0 3
#> Bbs1 1 2 0 4 0
dim(Rep11_MOB_0)
#> [1] 16218 262The MOB_sample_info object contains a data.frame with coordinates for each
spot.
head(MOB_sample_info)3.1.1 Filter out pratically unobserved genes
Rep11_MOB_0 <- Rep11_MOB_0[rowSums(Rep11_MOB_0) >= 3, ]3.1.2 Get total_counts for every spot
Rep11_MOB_0 <- Rep11_MOB_0[, row.names(MOB_sample_info)]
MOB_sample_info$total_counts <- colSums(Rep11_MOB_0)
head(MOB_sample_info)3.1.3 Get coordinates from MOB_sample_info
X <- MOB_sample_info[, c("x", "y")]
head(X)3.2 stabilize
The SpatialDE method assumes normally distributed data, so we stabilize the variance of the negative binomial distributed counts data using Anscombe’s approximation.
The stabilize() function takes as input a data.frame of expression values with samples in columns and genes in rows. Thus, in this case, we have to transpose the data.
norm_expr <- stabilize(Rep11_MOB_0)
#> + '/home/biocbuild/.cache/R/basilisk/1.12.0/0/bin/conda' 'create' '--yes' '--prefix' '/home/biocbuild/.cache/R/basilisk/1.12.0/spatialDE/1.6.0/env' 'python=3.8.13' '--quiet' '-c' 'conda-forge'
#> + '/home/biocbuild/.cache/R/basilisk/1.12.0/0/bin/conda' 'install' '--yes' '--prefix' '/home/biocbuild/.cache/R/basilisk/1.12.0/spatialDE/1.6.0/env' 'python=3.8.13'
#> + '/home/biocbuild/.cache/R/basilisk/1.12.0/0/bin/conda' 'install' '--yes' '--prefix' '/home/biocbuild/.cache/R/basilisk/1.12.0/spatialDE/1.6.0/env' '-c' 'conda-forge' 'python=3.8.13' 'numpy=1.23.5' 'patsy=0.5.3' 'pandas=1.5.2'
norm_expr[1:5, 1:5]
#> 16.92x9.015 16.945x11.075 16.97x10.118 16.939x12.132 16.949x13.055
#> Nrf1 1.227749 0.8810934 0.8810934 1.2277491 0.8810934
#> Zbtb5 1.227749 0.8810934 1.2277491 0.8810934 0.8810934
#> Ccnl1 1.227749 1.6889027 1.2277491 1.2277491 0.8810934
#> Lrrfip1 1.484676 1.4846765 0.8810934 0.8810934 1.6889027
#> Bbs1 1.227749 1.4846765 0.8810934 1.8584110 0.88109343.3 regress_out
Next, we account for differences in library size between the samples by regressing out the effect of the total counts for each gene using linear regression.
resid_expr <- regress_out(norm_expr, sample_info = MOB_sample_info)
resid_expr[1:5, 1:5]
#> 16.92x9.015 16.945x11.075 16.97x10.118 16.939x12.132 16.949x13.055
#> Nrf1 -0.75226761 -1.2352000 -1.0164479 -0.7903289 -1.0973214
#> Zbtb5 0.09242373 -0.3323719 0.1397144 -0.2760560 -0.2533134
#> Ccnl1 -2.77597164 -2.5903783 -2.6092013 -2.8529340 -3.1193883
#> Lrrfip1 -1.92331333 -2.1578718 -2.3849405 -2.5924072 -1.7163300
#> Bbs1 -1.12186064 -1.0266476 -1.3706460 -0.5363646 -1.46661553.4 run
To reduce running time, the SpatialDE test is run on a subset of 1000 genes. Running it on the complete data set takes about 10 minutes.
# For this example, run spatialDE on the first 1000 genes
sample_resid_expr <- head(resid_expr, 1000)
results <- spatialDE::run(sample_resid_expr, coordinates = X)
head(results[order(results$qval), ])3.5 model_search
Finally, we can classify the DE genes to interpetable DE classes using the model_search function.
We apply the model search on filtered DE results, using a threshold of 0.05 for the Q-values.
de_results <- results[results$qval < 0.05, ]
ms_results <- model_search(
sample_resid_expr,
coordinates = X,
de_results = de_results
)
# To show ms_results sorted on qvalue, uncomment the following line
# head(ms_results[order(ms_results$qval), ])
head(ms_results)3.6 spatial_patterns
Furthermore, we can group spatially variable genes (SVGs) into spatial patterns using automatic expression histology (AEH).
sp <- spatial_patterns(
sample_resid_expr,
coordinates = X,
de_results = de_results,
n_patterns = 4L, length = 1.5
)
sp$pattern_results3.7 Plots
Visualizing one of the most significant genes.
gene <- "Pcp4"
ggplot(data = MOB_sample_info, aes(x = x, y = y, color = norm_expr[gene, ])) +
geom_point(size = 7) +
ggtitle(gene) +
scale_color_viridis_c() +
labs(color = gene)
3.7.1 Plot Spatial Patterns of Multiple Genes
As an alternative to the previous figure, we can plot multiple genes using the normalized expression values.
ordered_de_results <- de_results[order(de_results$qval), ]
multiGenePlots(norm_expr,
coordinates = X,
ordered_de_results[1:6, ]$g,
point_size = 4,
viridis_option = "D",
dark_theme = FALSE
)
3.8 Plot Fraction Spatial Variance vs Q-value
FSV_sig(results, ms_results)
#> Warning: ggrepel: 8 unlabeled data points (too many overlaps). Consider
#> increasing max.overlaps
4 SpatialExperiment integration
The SpatialDE workflow can also be executed with a SpatialExperiment object as input.
library(SpatialExperiment)
# For SpatialExperiment object, we neeed to transpose the counts matrix in order
# to have genes on rows and spot on columns.
# For this example, run spatialDE on the first 1000 genes
partial_counts <- head(Rep11_MOB_0, 1000)
spe <- SpatialExperiment(
assays = list(counts = partial_counts),
spatialData = DataFrame(MOB_sample_info[, c("x", "y")]),
spatialCoordsNames = c("x", "y")
)
out <- spatialDE(spe, assay_type = "counts", verbose = FALSE)
head(out[order(out$qval), ])4.1 Plot Spatial Patterns of Multiple Genes (using SpatialExperiment object)
We can plot spatial patterns of multiples genes using the spe object.
spe_results <- out[out$qval < 0.05, ]
ordered_spe_results <- spe_results[order(spe_results$qval), ]
multiGenePlots(spe,
assay_type = "counts",
ordered_spe_results[1:6, ]$g,
point_size = 4,
viridis_option = "D",
dark_theme = FALSE
)
4.2 Classify spatially variable genes with model_search and spatial_patterns
msearch <- modelSearch(spe,
de_results = out, qval_thresh = 0.05,
verbose = FALSE
)
head(msearch)spatterns <- spatialPatterns(spe,
de_results = de_results, qval_thresh = 0.05,
n_patterns = 4L, length = 1.5, verbose = FALSE
)
spatterns$pattern_resultssessionInfo
Session info
#> [1] "2023-04-25 18:56:43 EDT"
#> R version 4.3.0 RC (2023-04-13 r84269)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: Ubuntu 22.04.2 LTS
#>
#> Matrix products: default
#> BLAS: /home/biocbuild/bbs-3.17-bioc/R/lib/libRblas.so
#> LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_GB 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: America/New_York
#> tzcode source: system (glibc)
#>
#> attached base packages:
#> [1] stats4 stats graphics grDevices utils datasets methods
#> [8] base
#>
#> other attached packages:
#> [1] SpatialExperiment_1.10.0 SingleCellExperiment_1.22.0
#> [3] SummarizedExperiment_1.30.0 Biobase_2.60.0
#> [5] GenomicRanges_1.52.0 GenomeInfoDb_1.36.0
#> [7] IRanges_2.34.0 S4Vectors_0.38.0
#> [9] BiocGenerics_0.46.0 MatrixGenerics_1.12.0
#> [11] matrixStats_0.63.0 ggplot2_3.4.2
#> [13] spatialDE_1.6.0 BiocStyle_2.28.0
#>
#> loaded via a namespace (and not attached):
#> [1] tidyselect_1.2.0 viridisLite_0.4.1
#> [3] farver_2.1.1 dplyr_1.1.2
#> [5] filelock_1.0.2 R.utils_2.12.2
#> [7] bitops_1.0-7 fastmap_1.1.1
#> [9] RCurl_1.98-1.12 digest_0.6.31
#> [11] lifecycle_1.0.3 magrittr_2.0.3
#> [13] compiler_4.3.0 rlang_1.1.0
#> [15] sass_0.4.5 tools_4.3.0
#> [17] utf8_1.2.3 yaml_2.3.7
#> [19] knitr_1.42 labeling_0.4.2
#> [21] dqrng_0.3.0 reticulate_1.28
#> [23] here_1.0.1 DelayedArray_0.26.0
#> [25] BiocParallel_1.34.0 HDF5Array_1.28.0
#> [27] withr_2.5.0 R.oo_1.25.0
#> [29] grid_4.3.0 fansi_1.0.4
#> [31] beachmat_2.16.0 colorspace_2.1-0
#> [33] Rhdf5lib_1.22.0 edgeR_3.42.0
#> [35] scales_1.2.1 cli_3.6.1
#> [37] rmarkdown_2.21 generics_0.1.3
#> [39] rjson_0.2.21 DelayedMatrixStats_1.22.0
#> [41] scuttle_1.10.0 cachem_1.0.7
#> [43] rhdf5_2.44.0 zlibbioc_1.46.0
#> [45] parallel_4.3.0 BiocManager_1.30.20
#> [47] XVector_0.40.0 basilisk_1.12.0
#> [49] vctrs_0.6.2 Matrix_1.5-4
#> [51] jsonlite_1.8.4 dir.expiry_1.8.0
#> [53] bookdown_0.33 ggrepel_0.9.3
#> [55] magick_2.7.4 locfit_1.5-9.7
#> [57] limma_3.56.0 jquerylib_0.1.4
#> [59] glue_1.6.2 DropletUtils_1.20.0
#> [61] codetools_0.2-19 gtable_0.3.3
#> [63] munsell_0.5.0 tibble_3.2.1
#> [65] pillar_1.9.0 basilisk.utils_1.12.0
#> [67] htmltools_0.5.5 rhdf5filters_1.12.0
#> [69] GenomeInfoDbData_1.2.10 R6_2.5.1
#> [71] sparseMatrixStats_1.12.0 rprojroot_2.0.3
#> [73] evaluate_0.20 lattice_0.21-8
#> [75] highr_0.10 png_0.1-8
#> [77] R.methodsS3_1.8.2 backports_1.4.1
#> [79] bslib_0.4.2 Rcpp_1.0.10
#> [81] gridExtra_2.3 checkmate_2.1.0
#> [83] xfun_0.39 pkgconfig_2.0.3