Overview
This vignette demonstrates the use of the SpatialDecon package to estimate cell abundance in spatial gene expression studies.
We’ll analyze a dense GeoMx dataset from a lung tumor, looking for abundance of different immune cell types. This dataset has 30 ROIs. In each ROI, Tumor and Microenvironment segments have been profiled separately.
Data preparation
First, we load the package:
library(SpatialDecon)
library(GeomxTools)
#> Loading required package: Biobase
#> Loading required package: BiocGenerics
#>
#> Attaching package: 'BiocGenerics'
#> The following objects are masked from 'package:stats':
#>
#> IQR, mad, sd, var, xtabs
#> The following objects are masked from 'package:base':
#>
#> Filter, Find, Map, Position, Reduce, anyDuplicated, append,
#> as.data.frame, basename, cbind, colnames, dirname, do.call,
#> duplicated, eval, evalq, get, grep, grepl, intersect, is.unsorted,
#> lapply, mapply, match, mget, order, paste, pmax, pmax.int, pmin,
#> pmin.int, rank, rbind, rownames, sapply, setdiff, sort, table,
#> tapply, union, unique, unsplit, which.max, which.min
#> Welcome to Bioconductor
#>
#> Vignettes contain introductory material; view with
#> 'browseVignettes()'. To cite Bioconductor, see
#> 'citation("Biobase")', and for packages 'citation("pkgname")'.
#> Loading required package: NanoStringNCTools
#> Loading required package: S4Vectors
#> Loading required package: stats4
#>
#> Attaching package: 'S4Vectors'
#> The following objects are masked from 'package:base':
#>
#> I, expand.grid, unname
#> Loading required package: ggplot2
Now let’s load our example data and examine it:
The spatialdecon function takes 3 arguments of expression data:
- The normalized data.
- A matrix of expected background for all data points in the normalized data matrix.
- Optionally, either a matrix of per-data-point weights, or the raw data, which is used to derive weights (low counts are less statistically stable, and this allows spatialdecon to down-weight them.)
We estimate each data point’s expected background from the negative control probes from its corresponding observation: This study has two probesets, and the genes from each probeset will have distinct background values. The function “derive_GeoMx_background” conveniently estimates background of all data points, accounting for which probeset each gene belongs to:
Cell profile matrices
A “cell profile matrix” is a pre-defined matrix that specifies the expected expression profiles of each cell type in the experiment. The SpatialDecon library comes with one such matrix pre-loaded, the “SafeTME” matrix, designed for estimation of immune and stroma cells in the tumor microenvironment. (This matrix was designed to avoid genes commonly expressed by cancer cells; see the SpatialDecon manuscript for details.)
Let’s take a glance at the safeTME matrix:
data("safeTME")
data("safeTME.matches")
signif(safeTME[seq_len(3), seq_len(3)], 2)
#> macrophages mast B.naive
#> A2M 0.8500 0.044 0.0043
#> ABCB1 0.0021 0.023 0.0250
#> ABCB4 0.0044 0.000 0.2200
heatmap(sweep(safeTME, 1, apply(safeTME, 1, max), "/"),
labRow = NA, margins = c(10, 5))
For studies of other tissue types, we have provided a library of cell profile matrices, available on Github and downloadable with the “download_profile_matrix” function.
For a complete list of matrices, see CellProfileLibrary GitHub Page.
Below we download a matrix of cell profiles derived from scRNA-seq of a mouse spleen.
mousespleen <- download_profile_matrix(species = "Mouse",
age_group = "Adult",
matrixname = "Spleen_MCA")
dim(mousespleen)
#> [1] 11125 9
mousespleen[1:4,1:4]
#> Dendritic.cell.S100a4.high Dendritic.cell.Siglech.high
#> 0610009B22Rik 0.02985075 0.0000000
#> 0610010F05Rik 0.00000000 0.0000000
#> 0610010K14Rik 0.02985075 0.0000000
#> 0610012G03Rik 0.08955224 0.1111111
#> Granulocyte Macrophage
#> 0610009B22Rik 0.00000000 0.00000000
#> 0610010F05Rik 0.00000000 0.00000000
#> 0610010K14Rik 0.00000000 0.03846154
#> 0610012G03Rik 0.08571429 0.03846154
head(cellGroups)
#> $Dendritic
#> [1] "Dendritic.cell.S100a4.high" "Dendritic.cell.Siglech.high"
#>
#> $Granulocyte
#> [1] "Granulocyte"
#>
#> $Macrophage
#> [1] "Macrophage"
#>
#> $`Marginal zone B`
#> [1] "Marginal.zone.B.cell"
#>
#> $Monocyte
#> [1] "Monocyte"
#>
#> $NK
#> [1] "NK.cell"
metadata
#> Profile Matrix Tissue Species Strain Age Age Group
#> 33 MCA Spleen Mouse C57BL/6 6-10 weeks Adult
#> URL
#> 33 https://pubmed.ncbi.nlm.nih.gov/29474909/
#> Citation
#> 33 Han, X. et al. Mapping the Mouse Cell Atlas by Microwell-Seq. Cell172, 1091-1107.e17 (2018).
heatmap(sweep(mousespleen, 1, apply(mousespleen, 1, max), "/"),
labRow = NA, margins = c(10, 5), cexCol = 0.7)
For studies where the provided cell profile matrices aren’t sufficient or if a specific single cell dataset is wanted, we can make a custom profile matrix using the function create_profile_matrix().
This mini single cell dataset is a fraction of the data from Kinchen, J. et al. Structural Remodeling of the Human Colonic Mesenchyme in Inflammatory Bowel Disease. Cell 175, 372-386.e17 (2018).
Pericyte cell and smooth muscle cell of colon will be dropped from this matrix due to low cell count. The average expression across all cells of one type is returned so the more cells of one type, the better reflection of the true gene expression. The confidence in these averages can be changed using the minCellNum filter.
custom_mtx <- create_profile_matrix(mtx = mini_singleCell_dataset$mtx, # cell x gene count matrix
cellAnnots = mini_singleCell_dataset$annots, # cell annotations with cell type and cell name as columns
cellTypeCol = "LabeledCellType", # column containing cell type
cellNameCol = "CellID", # column containing cell ID/name
matrixName = "custom_mini_colon", # name of final profile matrix
outDir = NULL, # path to desired output directory, set to NULL if matrix should not be written
normalize = FALSE, # Should data be normalized?
minCellNum = 5, # minimum number of cells of one type needed to create profile, exclusive
minGenes = 10, # minimum number of genes expressed in a cell, exclusive
scalingFactor = 5, # what should all values be multiplied by for final matrix
discardCellTypes = TRUE) # should cell types be filtered for types like mitotic, doublet, low quality, unknown, etc.
#> [1] "Creating Atlas"
#> [1] "1 / 6 : stromal cell"
#> [1] "2 / 6 : glial cell"
#> [1] "3 / 6 : endothelial cell"
#> [1] "4 / 6 : plasma cell"
#> [1] "5 / 6 : pericyte cell"
#> Warning in create_profile_matrix(mtx = mini_singleCell_dataset$mtx, cellAnnots = mini_singleCell_dataset$annots, :
#> pericyte cell was dropped from matrix because it didn't have enough viable cells based on current filtering thresholds.
#> If this cell type is necessary consider changing minCellNum or minGenes
#> [1] "6 / 6 : smooth muscle cell of colon"
#> Warning in create_profile_matrix(mtx = mini_singleCell_dataset$mtx, cellAnnots = mini_singleCell_dataset$annots, :
#> smooth muscle cell of colon was dropped from matrix because it didn't have enough viable cells based on current filtering thresholds.
#> If this cell type is necessary consider changing minCellNum or minGenes
head(custom_mtx)
#> stromal cell glial cell endothelial cell plasma cell
#> PLEKHN1 0.0000000 0 0.0000000 0.05787067
#> PERM1 0.1167131 0 0.0000000 0.00000000
#> C1orf159 0.0000000 0 0.0000000 0.07922668
#> TTLL10 0.0000000 0 0.1853931 0.00000000
#> TAS1R3 0.0000000 0 0.0000000 0.07039008
#> ATAD3C 0.1219237 0 0.0000000 0.07922668
heatmap(sweep(custom_mtx, 1, apply(custom_mtx, 1, max), "/"),
labRow = NA, margins = c(10, 5), cexCol = 0.7)
Using the advanced settings of spatialdecon
spatialdecon has several abilities beyond basic deconvolution:
- If given the nuclei counts for each region/observation, it returns results on the scale of total cell counts.
- If given the identities of pure tumor regions/observations, it infers a handful of tumor-specific expression profiles and appends them to the cell profile matrix. Doing this accounts for cancer cell-derived expression from any genes in the cell profile matrix, removing contaminating signal from cancer cells.
- If given raw count data, it derives per-data-point weights, using an error model derived for GeoMx data.
- If given a “cellmatches” argument, it sums multiple closely-related cell types into a single score. E.g. if the safeTME matrix is used with the cell-matching data object “safeTME.matches”, it e.g. sums the “T.CD8.naive” and “T.CD8.memory” scores into a single “CD8.T.cells” score.
Let’s take a look at an example cell matching object:
Now let’s run spatialdecon:
# vector identifying pure tumor segments:
nsclc$istumor = nsclc$AOI.name == "Tumor"
# run spatialdecon with all the bells and whistles:
restils = runspatialdecon(object = nsclc,
norm_elt = "exprs_norm", # normalized data
raw_elt = "exprs", # expected background counts for every data point in norm
X = safeTME, # safeTME matrix, used by default
cellmerges = safeTME.matches, # safeTME.matches object, used by default
cell_counts = nsclc$nuclei, # nuclei counts, used to estimate total cells
is_pure_tumor = nsclc$istumor, # identities of the Tumor segments/observations
n_tumor_clusters = 5) # how many distinct tumor profiles to append to safeTME
str(pData(restils))
#> 'data.frame': 199 obs. of 21 variables:
#> $ Sample_ID : chr "ICP20th-L11-ICPKilo-ROI01-Tumor-A02" "ICP20th-L11-ICPKilo-ROI01-TME-A03" "ICP20th-L11-ICPKilo-ROI02-Tumor-A04" "ICP20th-L11-ICPKilo-ROI02-TME-A05" ...
#> $ Tissue : chr "L11" "L11" "L11" "L11" ...
#> $ Slide.name : chr "ICPKilo" "ICPKilo" "ICPKilo" "ICPKilo" ...
#> $ ROI : chr "ROI01" "ROI01" "ROI02" "ROI02" ...
#> $ AOI.name : chr "Tumor" "TME" "Tumor" "TME" ...
#> $ AOI.annotation : chr "PanCK" "TME" "PanCK" "TME" ...
#> $ x : int 0 0 600 600 1200 1200 1800 1800 2400 2400 ...
#> $ y : num 8000 8000 8000 8000 8000 8000 8000 8000 8000 8000 ...
#> $ nuclei : int 572 733 307 697 583 484 506 706 851 552 ...
#> $ istumor : logi TRUE FALSE TRUE FALSE TRUE FALSE ...
#> $ beta : num [1:199, 1:14] 0.4668 2.7947 0.9003 10.2149 0.0131 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ p : num [1:199, 1:14] 2.88e-01 7.34e-04 1.20e-01 3.84e-10 9.63e-01 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ t : num [1:199, 1:14] 1.0616 3.3765 1.5565 6.2603 0.0466 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ se : num [1:199, 1:14] 0.44 0.828 0.578 1.632 0.281 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ prop_of_all : num [1:199, 1:14] 0.2979 0.1734 0.4577 0.4794 0.0763 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ prop_of_nontumor : num [1:199, 1:14] 0.2979 0.1734 0.4577 0.4794 0.0763 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ sigma : num [1:199, 1:14, 1:14] 0.1933 0.6851 0.3346 2.6624 0.0788 ...
#> ..- attr(*, "dimnames")=List of 3
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:14] "var1" "var2" "var3" "var4" ...
#> .. ..$ : chr [1:14] "var1" "var2" "var3" "var4" ...
#> $ beta.granular : num [1:199, 1:23] 0.4668 2.7947 0.9003 10.2149 0.0131 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:23] "macrophages" "mast" "B.naive" "B.memory" ...
#> $ sigma.granular : num [1:199, 1:23, 1:23] 0.1933 0.6851 0.3346 2.6624 0.0788 ...
#> ..- attr(*, "dimnames")=List of 3
#> .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. ..$ : chr [1:23] "var1" "var2" "var3" "var4" ...
#> .. ..$ : chr [1:23] "var1" "var2" "var3" "var4" ...
#> $ cell.counts :'data.frame': 199 obs. of 2 variables:
#> ..$ cells.per.100: num [1:199, 1:14] 1.1831 7.0835 2.282 25.8905 0.0332 ...
#> .. ..- attr(*, "dimnames")=List of 2
#> .. .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> ..$ cell.counts : num [1:199, 1:14] 6.767 51.922 7.006 180.457 0.193 ...
#> .. ..- attr(*, "dimnames")=List of 2
#> .. .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. .. ..$ : chr [1:14] "macrophages" "mast" "B" "plasma" ...
#> $ cell.counts.granular:'data.frame': 199 obs. of 2 variables:
#> ..$ cells.per.100: num [1:199, 1:18] 1.1831 7.0835 2.282 25.8905 0.0332 ...
#> .. ..- attr(*, "dimnames")=List of 2
#> .. .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. .. ..$ : chr [1:18] "macrophages" "mast" "B.naive" "B.memory" ...
#> ..$ cell.counts : num [1:199, 1:18] 6.767 51.922 7.006 180.457 0.193 ...
#> .. ..- attr(*, "dimnames")=List of 2
#> .. .. ..$ : chr [1:199] "ROI01Tumor" "ROI01TME" "ROI02Tumor" "ROI02TME" ...
#> .. .. ..$ : chr [1:18] "macrophages" "mast" "B.naive" "B.memory" ...
names(restils@assayData)
#> [1] "exprs_norm" "resids" "exprs" "yhat"
str(restils@experimentData@other)
#> List of 1
#> $ SpatialDeconMatrix: num [1:544, 1:23] 0.0137 0.0379 0.2697 0.2754 0.4022 ...
#> ..- attr(*, "dimnames")=List of 2
#> .. ..$ : chr [1:544] "ADAM12" "ANGPTL4" "BCAT1" "BCL2A1" ...
#> .. ..$ : chr [1:23] "macrophages" "mast" "B.naive" "B.memory" ...
There are quite a few readouts here. Let’s review the important ones:
- beta: the cell abundance scores of the rolled-up/major cell types
- beta.granular: the cell abundance scores of the granular cell types, corresponding to the columns of the cell profile matrix
- yhat, resids: the fitted values and log2-scale residuals from the deconvolution fit. Can be used to measure each observation’s goodness-of-fit, a possible QC metric.
- cell.counts, cell.counts.granular: estimated numbers of each cell type, derived using the nuclei count input
- prop_of_nontumor: the beta matrix rescaled to give the proportions of non-tumor cells in each observation.
- SpatialDeconMatrix: the cell profile matrix used, including newly-derived tumor-specific columns.
To illustrate the derivation of tumor profiles, let’s look at the cell profile matrix output by spatialdecon:
Note the new tumor-specific columns.
Plotting deconvolution results
The SpatialDecon package contains two specialized plotting functions, and a default color palette for the safeTME matrix.
The first function is “TIL_barplot”, which is just a convenient way of drawing barplots of cell type abundance.
The second function is “florets”, used for plotting cell abundances atop some 2-D projection. Here, we’ll plot cell abundances atop the first 2 principal components of the data:
# PCA of the normalized data:
pc = prcomp(t(log2(pmax(nsclc@assayData$exprs_norm, 1))))$x[, c(1, 2)]
# run florets function:
par(mar = c(5,5,1,1))
layout(mat = (matrix(c(1, 2), 1)), widths = c(6, 2))
florets(x = pc[, 1], y = pc[, 2],
b = t(restils$beta), cex = .5,
xlab = "PC1", ylab = "PC2")
par(mar = c(0,0,0,0))
frame()
legend("center", fill = cellcols[colnames(restils$beta)],
legend = colnames(restils$beta), cex = 0.7)
So we can see that PC1 roughly tracks many vs. few immune cells, and PC2 tracks the relative abundance of lymphoid/myeloid populations.
Other functions
The SpatialDecon library includes several helpful functions for further analysis/fine-tuning of deconvolution results.
Combining cell types:
When two cell types are too similar, the estimation of their abundances becomes unstable. However, their sum can still be estimated easily. The function “collapseCellTypes” takes a deconvolution results object and collapses any closely-related cell types you tell it to:
Inferring an expression profile for a cell type omitted from the profile matrix
Sometimes a cell profile matrix will omit a cell type you know to be present in a tissue. If your data includes any regions that are purely this unmodelled cell type - e.g. because you’ve used the GeoMx platform’s segmentation capability to specifically select them - then you can infer a profile for that cell type and merge it with your cell profile matrix. The algorithm clusters all the observations you designate as purely the unmodelled cell type, and collapses those clusters into as many profiles of that cell type as you wish. For cancer cell, it may be appropriate to specify 10 or more clusters; for highly regular healthy cells, one cluster may suffice.
(Note: this functionality can also be run within the spatialdecon function, as is demonstrated further above.)
pure.tumor.ids = res$AOI.name == "Tumor"
safeTME.with.tumor = runMergeTumorIntoX(object = nsclc,
norm_elt = "exprs_norm",
pure_tumor_ids = pure.tumor.ids,
X = safeTME,
K = 3)
heatmap(sweep(safeTME.with.tumor, 1, apply(safeTME.with.tumor, 1, max), "/"),
labRow = NA, margins = c(10, 5))
Reverse deconvolution
Once cell type abundance has been estimated, we can flip the deconvolution around, modelling the expression data as a function of cell abundances, and thereby deriving:
- Estimated expression of each gene in each cell type. (Including for genes not present in your cell profile matrix)
- Fitted expression values for each gene based on cell mixing.
- Residuals of each gene: how does their expression compare to what cell mixing would predict?
- Two metrics of how well genes are predicted by/ redundant with cell mixing: correlation between observed and fitted expression, and residual SD.
The function “reversedecon” runs this model.
# look at the two metrics of goodness-of-fit:
plot(fData(rdecon)$cors, fData(rdecon)$resid.sd, col = 0)
showgenes = c("CXCL14", "LYZ", "FOXP3")
text(fData(rdecon)$cors[!rownames(fData(rdecon)) %in% showgenes],
fData(rdecon)$resid.sd[!rownames(fData(rdecon)) %in% showgenes],
setdiff(rownames(fData(rdecon)), showgenes), cex = 0.5)
text(fData(rdecon)$cors[rownames(fData(rdecon)) %in% showgenes], fData(rdecon)$resid.sd[rownames(fData(rdecon)) %in% showgenes],
showgenes, cex = 0.75, col = 2)
From the above plot, we can see that genes like CXCL14 vary independently of cell mixing, genes like LYZ are correlated with cell mixing but still have variable expression, and genes like FOXP3 serve as nothing but obtuse readouts of cell mixing.
