DESpace can also be used to reveal the specific areas of the tissue affected by SVGs; i.e., spatial clusters that are particularly over/under abundant compared to the average signal. This function can be used to identify SVGs for each individual cluster.
Usage
individual_svg(
spe,
cluster_col,
sample_col = "sample_id",
edgeR_y = NULL,
min_counts = 20,
min_non_zero_spots = 10,
filter_gene = TRUE,
replicates = FALSE,
BPPARAM = NULL
)Arguments
- spe
SpatialExperiment or SingleCellExperiment.
- cluster_col
Column name of spatial clusters in
colData(spe).- sample_col
Column name of sample ids in
colData(spe).- edgeR_y
Pre-estimated dispersion; if it's null, compute dispersion.
- min_counts
Minimum number of counts per sample (across all spots) for a gene to be analyzed.
- min_non_zero_spots
Minimum number of non-zero spots per sample, for a gene to be analyzed.
- filter_gene
A logical. If TRUE,
svg_testfilters genes: genes have to be expressed in at least 'min_non_zero_spots' spots, and a gene requires at least 'min counts' counts per sample (across all locations).- replicates
Single sample or multi-sample test.
- BPPARAM
An optional parameter passed internally to bplapply. We suggest using as many cores as the number of spatial clusters. If unspecified, the script does not run in parallel. Note that parallel coding performs better only when dispersion estimations are not provided beforehand. Moreover, parallelizing the script will increase the memory requirement; if memory is an issue, leave 'BPPARAM' unspecified and, hence, avoid parallelization.
Value
A list of results, with one result per spatial cluster in each element. Specifically, each item in the list is a "gene_results" dataframe which contains main edgeR test results.
Details
For every spatial cluster we test, edgeR would normally re-compute the dispersion estimates based on the specific design of the test.
However, this calculation represents the majority of the overall computing time.
Therefore, to speed-up calculations, we propose to use the dispersion estimates which were previously computed for the gene-level tests.
This introduces a minor approximation which, in our benchmarks, does not lead to decreased accuracy.
If you want to use pre-computed gene-level dispersion estimates, set edgeR_y to 'estimated_y'.
Alternatively, if you want to re-compute dispersion estimates (significantly slower, but marginally more accurate option), leave edgeR_y empty.
Examples
# load the input data:
data("LIBD_subset", package = "DESpace")
LIBD_subset
#> class: SpatialExperiment
#> dim: 500 3639
#> metadata(0):
#> assays(2): counts logcounts
#> rownames(500): ENSG00000282097 ENSG00000225129 ... ENSG00000267199
#> ENSG00000273129
#> rowData names(0):
#> colnames(3639): AAACAAGTATCTCCCA-1 AAACAATCTACTAGCA-1 ...
#> TTGTTTGTATTACACG-1 TTGTTTGTGTAAATTC-1
#> colData names(4): array_row array_col layer_guess_reordered sample_id
#> reducedDimNames(0):
#> mainExpName: NULL
#> altExpNames(0):
#> spatialCoords names(0) :
#> imgData names(0):
# load pre-computed results (obtaines via `svg_test`)
data("results_svg_test", package = "DESpace")
# svg_test returns of a list of 2 objects:
# "gene_results": a dataframe contains main edgeR test results;
# "estimated_y": a DGEList object contains the estimated common dispersion,
# which can later be used to speed-up calculation when testing individual clusters.
# We visualize differential results:
head(results_svg_test$gene_results, 3)
#> gene_id LR logCPM PValue FDR
#> ENSG00000115756 ENSG00000115756 1073.3689 15.89723 1.204871e-228 2.855545e-226
#> ENSG00000054690 ENSG00000054690 1056.7744 15.48088 4.686692e-225 5.553730e-223
#> ENSG00000111716 ENSG00000111716 752.4418 16.58993 2.894030e-159 2.286283e-157
# Individual cluster test: identify SVGs for each individual cluster
# set parallel computing; we suggest using as many cores as the number of spatial clusters.
# Note that parallelizing the script will increase the memory requirement;
# if memory is an issue, leave 'BPPARAM' unspecified and, hence, avoid parallelization.
set.seed(123)
results_individual_svg <- individual_svg(LIBD_subset,
edgeR_y = results_svg_test$estimated_y,
cluster_col = "layer_guess_reordered")
#> Filter low quality genes:
#> min_counts = 20; min_non_zero_spots = 10.
#> The number of genes that pass filtering is237.
#> Pre-processing
#> Start modeling
#> Returning results
# We visualize results for the cluster 'WM'
results_WM <- results_individual_svg[[7]]
head(results_WM,3)
#> gene_id LR logCPM logFC PValue
#> ENSG00000054690 ENSG00000054690 978.6555 15.48088 2.346996 7.831449e-215
#> ENSG00000166086 ENSG00000166086 572.4992 15.47657 1.845724 1.605722e-126
#> ENSG00000111716 ENSG00000111716 409.7061 16.58993 -1.079196 4.247092e-91
#> FDR
#> ENSG00000054690 1.856053e-212
#> ENSG00000166086 1.902780e-124
#> ENSG00000111716 3.355203e-89