Seurat::DotPlot() does not seem to be plotting the runEscape() results properly

[wfaalajr]

Hi! this tool has gotten a major facelift since I last used it.

I'm running the below code (as is with the vignette), but it seems plotting in DotPlot does not plot the enrichment values in the new assay created by runEscape(), any thoughts?

GS.hallmark <- getGeneSets(library = "H") #add the species = "Mus musculus" option here for mouse

enrichment.scores <- escape.matrix(deb.sct, 
                                   gene.sets = GS.hallmark, 
                                   groups = 5000, 
                                   min.size = 5)

deb.sct <- runEscape(deb.sct, 
                     method = "ssGSEA",
                     gene.sets = GS.hallmark, 
                     groups = 5000, 
                     min.size = 0,
                     new.assay.name = "escape.ssGSEA")

Seurat::DotPlot(deb.sct, features = c("HALLMARK-GLYCOLYSIS"), assay = "escape.ssGSEA") #seems not working

Is this an expected result or should I specify some other parameter for DotPlot()?

[ncborcherding]

@wfaalajr

First off - this is an excellent issue run down - thanks this helps me immensely when it comes to questions.

I reformatted your example slightly to use the built in data set in Seurat:

GS.hallmark <- getGeneSets(library = "H") #add the species = "Mus musculus" option here for mouse

pbmc_small <- runEscape(pbmc_small, 
                     method = "ssGSEA",
                     gene.sets = GS.hallmark, 
                     groups = 5000, 
                     min.size = 0,
                     new.assay.name = "escape.ssGSEA")

Seurat::DotPlot(pbmc_small, features = c("HALLMARK-GLYCOLYSIS"), assay = "escape.ssGSEA")

I get the following results for the plot:

It looks like there might not be any enrichment calculation that is performed. Can you give me more information on your experiment?

Also could you give me the output of sessionInfo() just so I know the system and version you are running.

Thanks,
Nick

[wfaalajr]

Hi! Thanks for the quick reply, this is a 10x single-cell data from human skin biopsies, the numbered cluster identities here correspond to skin cell types. I chose the glycolysis as this ought to be positive regardless of cell type (most of the time anyway). Funny enough I can actually plot using FeaturePlot() and VlnPlot() but not on DotPlot(), see resulting plots below:

FeaturePlot(deb.sct, "HALLMARK-GLYCOLYSIS",min.cutoff = 1500,order = T)

and

VlnPlot(deb.sct,features = c("HALLMARK-GLYCOLYSIS", "HALLMARK-APOPTOSIS"), assay = "escape.ssGSEA", stack = T, flip = T)

Showing the contents of the 'escape.ssGSEA' assay shows that the values are indeed there:

enrichment_data <- deb.sct[["escape.ssGSEA"]]$data
head(enrichment_data)

6 x 14307 sparse Matrix of class "dgCMatrix"
[[ suppressing 18 column names ‘AAACCCACACCTCGTT-1_1’, ‘AAACGCTGTGAAGCTG-1_1’, ‘AAAGAACGTTTGGAGG-1_1’ ... ]]

HALLMARK-ADIPOGENESIS 1910.0856 1859.442
HALLMARK-ALLOGRAFT-REJECTION 1827.6121 1898.668
HALLMARK-ANDROGEN-RESPONSE 1471.7648 1210.696
HALLMARK-ANGIOGENESIS 499.8108 -1045.376
HALLMARK-APICAL-JUNCTION 1370.9801 2018.616
HALLMARK-APICAL-SURFACE 2138.1332 1598.549

Lastly, my sessionInfo :)

sessionInfo()
R version 4.4.1 (2024-06-14)
Platform: aarch64-apple-darwin20
Running under: macOS 15.2

Matrix products: default
BLAS: /System/Library/Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/libBLAS.dylib
LAPACK: /Library/Frameworks/R.framework/Versions/4.4-arm64/Resources/lib/libRlapack.dylib; LAPACK version 3.12.0

locale:
[1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8

time zone: Asia/Manila
tzcode source: internal

attached base packages:
[1] stats graphics grDevices utils
[5] datasets methods base

other attached packages:
[1] ggplot2_3.5.1 Seurat_5.1.0
[3] SeuratObject_5.0.2 sp_2.1-4
[5] escape_2.2.2

loaded via a namespace (and not attached):
[1] igraph_2.0.3
[2] graph_1.80.0
[3] treemap_2.4-4
[4] ica_1.0-3
[5] plotly_4.10.4
[6] rematch2_2.1.2
[7] devtools_2.4.5
[8] zlibbioc_1.50.0
[9] fpc_2.2-12
[10] tidyselect_1.2.1
[11] bit_4.5.0
[12] doParallel_1.0.17
[13] lattice_0.22-6
[14] rjson_0.2.23
[15] factoextra_1.0.7
[16] blob_1.2.4
[17] stringr_1.5.1
[18] urlchecker_1.0.1
[19] S4Arrays_1.4.1
[20] parallel_4.4.1
[21] Biobase_2.64.0
[22] lemon_0.4.9
[23] png_0.1-8
[24] pathfindR.data_2.1.0
[25] cli_3.6.3
[26] ggplotify_0.1.2
[27] goftest_1.2-3
[28] kernlab_0.9-33
[29] purrr_1.0.2
[30] BiocNeighbors_1.20.2
[31] stringdist_0.9.12
[32] uwot_0.2.2
[33] curl_5.2.2
[34] mime_0.12
[35] evaluate_0.24.0
[36] tidytree_0.4.6
[37] leiden_0.4.3.1
[38] V8_6.0.0
[39] stringi_1.8.4
[40] backports_1.5.0
[41] desc_1.4.3
[42] XML_3.99-0.17
[43] lubridate_1.9.3
[44] httpuv_1.6.15
[45] flexmix_2.3-19
[46] AnnotationDbi_1.66.0
[47] paletteer_1.6.0
[48] magrittr_2.0.3
[49] scoper_1.3.0
[50] splines_4.4.1
[51] mclust_6.1.1
[52] ggraph_2.2.1
[53] dplyr_1.1.4
[54] sctransform_0.4.1
[55] ggbeeswarm_0.7.2
[56] sessioninfo_1.2.2
[57] HDF5Array_1.30.1
[58] DBI_1.2.3
[59] Nebulosa_1.14.0
[60] withr_3.0.1
[61] class_7.3-22
[62] lmtest_0.9-40
[63] GSEABase_1.64.0
[64] tidygraph_1.3.1
[65] immunarch_0.9.1
[66] BiocManager_1.30.25
[67] htmlwidgets_1.6.4
[68] fs_1.6.4
[69] S4Vectors_0.42.1
[70] SingleCellExperiment_1.26.0
[71] ggrepel_0.9.6
[72] labeling_0.4.3
[73] SparseArray_1.4.8
[74] cellranger_1.1.0
[75] DEoptimR_1.1-3
[76] MatrixGenerics_1.16.0
[77] diptest_0.77-1
[78] annotate_1.80.0
[79] reticulate_1.39.0
[80] zoo_1.8-12
[81] XVector_0.44.0
[82] knitr_1.48
[83] UCSC.utils_1.0.0
[84] BiocGenerics_0.50.0
[85] AUCell_1.24.0
[86] timechange_0.3.0
[87] foreach_1.5.2
[88] ggpointdensity_0.1.0
[89] fansi_1.0.6
[90] pathfindR_2.4.1
[91] patchwork_1.2.0
[92] grid_4.4.1
[93] rhdf5_2.46.1
[94] data.table_1.16.0
[95] ggtree_3.12.0
[96] R.oo_1.26.0
[97] RSpectra_0.16-2
[98] irlba_2.3.5.1
[99] ggrastr_1.0.2
[100] fastDummies_1.7.4
[101] gridGraphics_0.5-1
[102] ellipsis_0.3.2
[103] SpatialExperiment_1.14.0
[104] lazyeval_0.2.2
[105] yaml_2.3.10
[106] ade4_1.7-22
[107] survival_3.6-4
[108] scattermore_1.2
[109] crayon_1.5.3
[110] RcppAnnoy_0.0.22
[111] RColorBrewer_1.1-3
[112] tidyr_1.3.1
[113] progressr_0.14.0
[114] tweenr_2.0.3
[115] later_1.3.2
[116] ggridges_0.5.6
[117] codetools_0.2-20
[118] GlobalOptions_0.1.2
[119] profvis_0.3.8
[120] rlist_0.4.6.2
[121] KEGGREST_1.42.0
[122] Rtsne_0.17
[123] shape_1.4.6.1
[124] Rsamtools_2.20.0
[125] pkgconfig_2.0.3
[126] xml2_1.3.6
[127] spatstat.univar_3.0-1
[128] ggpubr_0.6.0
[129] GenomicAlignments_1.40.0
[130] GenomicRanges_1.56.1
[131] IRanges_2.38.1
[132] aplot_0.2.3
[133] spatstat.sparse_3.1-0
[134] ape_5.8
[135] viridisLite_0.4.2
[136] gridBase_0.4-7
[137] xtable_1.8-4
[138] car_3.1-2
[139] plyr_1.8.9
[140] httr_1.4.7
[141] tools_4.4.1
[142] globals_0.16.3
[143] prabclus_2.3-3
[144] pkgbuild_1.4.4
[145] beeswarm_0.4.0
[146] broom_1.0.6
[147] nlme_3.1-164
[148] digest_0.6.37
[149] Matrix_1.7-0
[150] farver_2.1.2
[151] tzdb_0.4.0
[152] dtplyr_1.3.1
[153] reshape2_1.4.4
[154] ks_1.14.2
[155] yulab.utils_0.1.7
[156] viridis_0.6.5
[157] glue_1.7.0
[158] cachem_1.1.0
[159] shazam_1.2.0
[160] polyclip_1.10-7
[161] UpSetR_1.4.0
[162] generics_0.1.3
[163] Biostrings_2.70.3
[164] stats4_4.4.1
[165] ggalluvial_0.12.5
[166] mvtnorm_1.3-1
[167] phylotate_1.3
[168] parallelly_1.38.0
[169] pkgload_1.4.0
[170] ggtext_0.1.2
[171] RcppHNSW_0.6.0
[172] ScaledMatrix_1.10.0
[173] carData_3.0-5
[174] pbapply_1.7-2
[175] randomcoloR_1.1.0.1
[176] SummarizedExperiment_1.34.0
[177] spam_2.10-0
[178] utf8_1.2.4
[179] seqinr_4.2-36
[180] graphlayouts_1.1.1
[181] readxl_1.4.3
[182] ggsignif_0.6.4
[183] gridExtra_2.3
[184] shiny_1.9.1
[185] GSVA_1.52.3
[186] GenomeInfoDbData_1.2.12
[187] rhdf5filters_1.14.1
[188] R.utils_2.12.3
[189] alakazam_1.3.0
[190] memoise_2.0.1
[191] rmarkdown_2.28
[192] pheatmap_1.0.12
[193] scales_1.3.0
[194] R.methodsS3_1.8.2
[195] future_1.34.0
[196] RANN_2.6.2
[197] spatstat.data_3.1-2
[198] rstudioapi_0.16.0
[199] cluster_2.1.6
[200] msigdbr_7.5.1
[201] janitor_2.2.0
[202] rsthemes_0.5.0
[203] spatstat.utils_3.1-0
[204] hms_1.1.3
[205] fitdistrplus_1.2-1
[206] munsell_0.5.1
[207] cowplot_1.1.3
[208] colorspace_2.1-1
[209] scCustomize_2.1.2
[210] rlang_1.1.4
[211] quadprog_1.5-8
[212] GenomeInfoDb_1.40.1
[213] DelayedMatrixStats_1.24.0
[214] sparseMatrixStats_1.14.0
[215] shinythemes_1.2.0
[216] dotCall64_1.1-1
[217] ggforce_0.4.2
[218] circlize_0.4.16
[219] xfun_0.47
[220] ggseqlogo_0.2
[221] remotes_2.5.0
[222] iterators_1.0.14
[223] matrixStats_1.4.1
[224] modeltools_0.2-23
[225] abind_1.4-8
[226] tibble_3.2.1
[227] treeio_1.28.0
[228] Rhdf5lib_1.24.2
[229] readr_2.1.5
[230] bitops_1.0-8
[231] ps_1.8.0
[232] promises_1.3.0
[233] RSQLite_2.3.7
[234] DelayedArray_0.30.1
[235] compiler_4.4.1
[236] forcats_1.0.0
[237] prettyunits_1.2.0
[238] distributional_0.5.0
[239] beachmat_2.18.1
[240] listenv_0.9.1
[241] Rcpp_1.0.13
[242] BiocSingular_1.18.0
[243] tensor_1.5
[244] usethis_3.0.0
[245] MASS_7.3-60.2
[246] progress_1.2.3
[247] UCell_2.6.2
[248] uuid_1.2-1
[249] BiocParallel_1.36.0
[250] gridtext_0.1.5
[251] babelgene_22.9
[252] spatstat.random_3.3-1
[253] R6_2.5.1
[254] fastmap_1.2.0
[255] fastmatch_1.1-4
[256] rstatix_0.7.2
[257] vipor_0.4.7
[258] airr_1.5.0
[259] dowser_2.2.0
[260] ROCR_1.0-11
[261] ggdist_3.3.2
[262] rsvd_1.0.5
[263] nnet_7.3-19
[264] gtable_0.3.5
[265] phangorn_2.12.1
[266] KernSmooth_2.23-24
[267] miniUI_0.1.1.1
[268] deldir_2.0-4
[269] htmltools_0.5.8.1
[270] bit64_4.5.2
[271] spatstat.explore_3.3-2
[272] lifecycle_1.0.4
[273] ggprism_1.0.5
[274] processx_3.8.4
[275] callr_3.7.6
[276] vctrs_0.6.5
[277] robustbase_0.99-4
[278] spatstat.geom_3.3-2
[279] snakecase_0.11.1
[280] ggfun_0.1.7
[281] sceasy_0.0.7
[282] future.apply_1.11.2
[283] pracma_2.4.4
[284] pillar_1.9.0
[285] magick_2.8.5
[286] jsonlite_1.8.8
[287] markdown_1.13

[ncborcherding]

@wfaalajr

Thanks so much for the follow up - I am glad to see the enrichment calculation itself is working.

This must be an issue with the handling of the escape assay by the DotPlot() function itself.

It looks like the following issue from Seurat might be relevant: satijalab/seurat#7719

I am going to close this issue as it is a Seurat compatibility issue and not an escape issue. But feel free to update if you get things to work.

Nick

[wfaalajr]

It seems to work when normalization was carried out using performNormalization(). I'm guessing the raw enrichment values are quite high and DotPlot expects scaled values. But running performNormalization() on an Assay5 (Seurat5 created objects) object will just return an error. So transferring the enrichment scores on the Assay5 Seuratv5 object as a new assay seems to work:

Idents(wt) <- wt$seurat_clusters
gene.sets <- getGeneSets(library = "H",species = "Mus musculus")
es_h5_ssgsea <- escape.matrix(input.data = wt,
                       method = "ssGSEA",
                       gene.sets = gene.sets)

#Transfer enrichment matrix to Seurat object as a new assay
new.assay <- CreateAssayObject(data = as.matrix(t(es_h5_ssgsea)))
wt[["escape.ssGSEA"]] <- new.assay

#Check if new assay looks good and numeric
max(wt[["escape.ssGSEA"]]$data)
min(wt[["escape.ssGSEA"]]$data)

#Normalize enrichment scores
wt <- performNormalization(wt, 
                           assay = "escape.ssGSEA", 
                           gene.sets = gene.sets)

#Check if new normalized enrichment assay looks good and numeric
max(wt[["escape.ssGSEA_normalized"]]$data)
min(wt[["escape.ssGSEA_normalized"]]$data)

#Visualizations
DotPlot(wt, features = c("HALLMARK-GLYCOLYSIS",
                         "HALLMARK-APOPTOSIS"),
        assay = "escape.ssGSEA_normalized",
        col.min = -8, col.max = 8) +
  ggplot2::coord_flip() +
  ggplot2::scale_colour_gradient2(low = "navy", mid = "#DCD6D0", high = "#94003a", midpoint = 0.5) +
  ggplot2::guides(color = ggplot2::guide_colorbar(title = "Pathway Activation")) +
  ggplot2::ylab("Cluster") +
  ggplot2::xlab("")

I got something like this from the above DotPlot code:

[ncborcherding]

It must be a value dependent phenomena like the Seurat issue suggests! Thanks so much for the follow up.

Nick