Modeling continuous cell-level covariates
Collapse using mean value for pseudobulk data
Run on 2023-06-16 15:31:53.141579
1 Introduction
Since read counts are summed across cells in a pseudobulk approach, modeling continuous cell-level covariates also requires a collapsing step. Here we summarize the values of a variable from a set of cells using the mean, and store the value for each cell type. Including these variables in a regression formula uses the summarized values from the corresponding cell type.
We demonstrate this feature on a lightly modified analysis of PBMCs from 8 individuals stimulated with interferon-β (Kang, et al, 2018, Nature Biotech).
2 Standard processing
Here is the code from the main vignette:
library(dreamlet)
library(muscat)
library(ExperimentHub)
library(scater)
# Download data, specifying EH2259 for the Kang, et al study
eh <- ExperimentHub()
sce <- eh[["EH2259"]]
# only keep singlet cells with sufficient reads
sce <- sce[rowSums(counts(sce) > 0) > 0, ]
sce <- sce[,colData(sce)$multiplets == 'singlet']
# compute QC metrics
qc <- perCellQCMetrics(sce)
# remove cells with few or many detected genes
ol <- isOutlier(metric = qc$detected, nmads = 2, log = TRUE)
sce <- sce[, !ol]
# set variable indicating stimulated (stim) or control (ctrl)
sce$StimStatus = sce$stimIn many datasets, continuous cell-level variables could be mapped reads, gene count, mitochondrial rate, etc. There are no continuous cell-level variables in this dataset, so we can simulate two from a normal distribution:
sce$value1 = rnorm(ncol(sce))
sce$value2 = rnorm(ncol(sce))3 Pseudobulk
Now compute the pseudobulk using standard code:
sce$id <- paste0(sce$StimStatus, sce$ind)
# Create pseudobulk
pb <- aggregateToPseudoBulk(sce,
assay = "counts",
cluster_id = "cell",
sample_id = "id",
verbose = FALSE)The means per variable, cell type, and sample are stored in the pseudobulk SingleCellExperiment object:
metadata(pb)$aggr_means## # A tibble: 128 × 5
## # Groups: cell [8]
## cell id cluster value1 value2
## <fct> <fct> <dbl> <dbl> <dbl>
## 1 B cells ctrl101 3.96 -0.0994 -0.00222
## 2 B cells ctrl1015 4.00 -0.0258 -0.0384
## 3 B cells ctrl1016 4 0.148 0.0666
## 4 B cells ctrl1039 4.04 -0.208 0.116
## 5 B cells ctrl107 4 -0.0559 -0.284
## 6 B cells ctrl1244 4 -0.138 -0.0202
## 7 B cells ctrl1256 4.01 -0.0550 -0.136
## 8 B cells ctrl1488 4.02 -0.0282 -0.0197
## 9 B cells stim101 4.09 -0.0224 -0.0622
## 10 B cells stim1015 4.06 0.0160 0.0462
## # ℹ 118 more rows4 Analysis
Including these variables in a regression formula uses the summarized values from the corresponding cell type. This happens behind the scenes, so the user doesn’t need to distinguish bewteen sample-level variables stored in colData(pb) and cell-level variables stored in metadata(pb)$aggr_means.
Variance partition and hypothesis testing proceeds as ususal:
form = ~ StimStatus + value1 + value2
# Normalize and apply voom/voomWithDreamWeights
res.proc = processAssays( pb, form, min.count=5)
# run variance partitioning analysis
vp.lst = fitVarPart( res.proc, form)
# Summarize variance fractions genome-wide for each cell type
plotVarPart(vp.lst, label.angle=60) 
# Differential expression analysis within each assay
res.dl = dreamlet( res.proc, form)
# dreamlet results include coefficients for value1 and value2
res.dl## class: dreamletResult
## assays(8): B cells CD14+ Monocytes ... Megakaryocytes NK cells
## Genes:
## min: 182
## max: 5262
## details(6): assay n_retain ... n_errors error_initial
## coefNames(4): (Intercept) StimStatusstim value1 value25 Session Info
## R version 4.3.0 (2023-04-21)
## Platform: x86_64-apple-darwin22.4.0 (64-bit)
## Running under: macOS Ventura 13.4
##
## Matrix products: default
## BLAS: /Users/gabrielhoffman/prog/R-4.3.0/lib/libRblas.dylib
## LAPACK: /usr/local/Cellar/r/4.3.0_1/lib/R/lib/libRlapack.dylib; LAPACK version 3.11.0
##
## 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: America/New_York
## tzcode source: internal
##
## attached base packages:
## [1] stats4 stats graphics grDevices datasets utils methods
## [8] base
##
## other attached packages:
## [1] muscData_1.14.0 scater_1.28.0
## [3] scuttle_1.10.1 SingleCellExperiment_1.22.0
## [5] SummarizedExperiment_1.30.1 Biobase_2.60.0
## [7] GenomicRanges_1.52.0 GenomeInfoDb_1.36.0
## [9] IRanges_2.34.0 S4Vectors_0.38.1
## [11] MatrixGenerics_1.12.0 matrixStats_1.0.0
## [13] ExperimentHub_2.8.0 AnnotationHub_3.8.0
## [15] BiocFileCache_2.8.0 dbplyr_2.3.2
## [17] BiocGenerics_0.46.0 muscat_1.14.0
## [19] dreamlet_0.99.14 variancePartition_1.31.6
## [21] BiocParallel_1.34.2 limma_3.56.2
## [23] ggplot2_3.4.2 BiocStyle_2.28.0
##
## loaded via a namespace (and not attached):
## [1] bitops_1.0-7 httr_1.4.6
## [3] RColorBrewer_1.1-3 doParallel_1.0.17
## [5] Rgraphviz_2.44.0 numDeriv_2016.8-1.1
## [7] tools_4.3.0 sctransform_0.3.5
## [9] backports_1.4.1 utf8_1.2.3
## [11] R6_2.5.1 GetoptLong_1.0.5
## [13] withr_2.5.0 prettyunits_1.1.1
## [15] gridExtra_2.3 cli_3.6.1
## [17] labeling_0.4.2 sass_0.4.6
## [19] KEGGgraph_1.60.0 SQUAREM_2021.1
## [21] mvtnorm_1.2-1 blme_1.0-5
## [23] mixsqp_0.3-48 zenith_1.2.0
## [25] parallelly_1.36.0 invgamma_1.1
## [27] RSQLite_2.3.1 generics_0.1.3
## [29] shape_1.4.6 gtools_3.9.4
## [31] dplyr_1.1.2 Matrix_1.5-4.1
## [33] ggbeeswarm_0.7.2 fansi_1.0.4
## [35] abind_1.4-5 lifecycle_1.0.3
## [37] yaml_2.3.7 edgeR_3.42.4
## [39] gplots_3.1.3 grid_4.3.0
## [41] blob_1.2.4 promises_1.2.0.1
## [43] crayon_1.5.2 lattice_0.21-8
## [45] beachmat_2.16.0 msigdbr_7.5.1
## [47] annotate_1.78.0 KEGGREST_1.40.0
## [49] magick_2.7.4 pillar_1.9.0
## [51] knitr_1.43 ComplexHeatmap_2.16.0
## [53] rjson_0.2.21 boot_1.3-28.1
## [55] corpcor_1.6.10 future.apply_1.11.0
## [57] codetools_0.2-19 glue_1.6.2
## [59] data.table_1.14.8 vctrs_0.6.2
## [61] png_0.1-8 Rdpack_2.4
## [63] gtable_0.3.3 assertthat_0.2.1
## [65] cachem_1.0.8 xfun_0.39
## [67] rbibutils_2.2.13 S4Arrays_1.0.4
## [69] mime_0.12 Rfast_2.0.7
## [71] iterators_1.0.14 ellipsis_0.3.2
## [73] interactiveDisplayBase_1.38.0 nlme_3.1-162
## [75] pbkrtest_0.5.2 bit64_4.0.5
## [77] progress_1.2.2 EnvStats_2.7.0
## [79] filelock_1.0.2 bslib_0.4.2
## [81] TMB_1.9.4 irlba_2.3.5.1
## [83] vipor_0.4.5 KernSmooth_2.23-21
## [85] colorspace_2.1-0 rmeta_3.0
## [87] DBI_1.1.3 DESeq2_1.40.1
## [89] tidyselect_1.2.0 bit_4.0.5
## [91] compiler_4.3.0 curl_5.0.0
## [93] graph_1.78.0 BiocNeighbors_1.18.0
## [95] DelayedArray_0.26.3 bookdown_0.34
## [97] scales_1.2.1 caTools_1.18.2
## [99] remaCor_0.0.14 rappdirs_0.3.3
## [101] stringr_1.5.0 digest_0.6.31
## [103] minqa_1.2.5 rmarkdown_2.22
## [105] aod_1.3.2 XVector_0.40.0
## [107] RhpcBLASctl_0.23-42 htmltools_0.5.5
## [109] pkgconfig_2.0.3 lme4_1.1-33
## [111] sparseMatrixStats_1.12.0 highr_0.10
## [113] mashr_0.2.69 fastmap_1.1.1
## [115] rlang_1.1.1 GlobalOptions_0.1.2
## [117] shiny_1.7.4 DelayedMatrixStats_1.22.0
## [119] farver_2.1.1 jquerylib_0.1.4
## [121] jsonlite_1.8.5 BiocSingular_1.16.0
## [123] RCurl_1.98-1.12 magrittr_2.0.3
## [125] GenomeInfoDbData_1.2.10 munsell_0.5.0
## [127] Rcpp_1.0.10 babelgene_22.9
## [129] viridis_0.6.3 EnrichmentBrowser_2.30.1
## [131] RcppZiggurat_0.1.6 stringi_1.7.12
## [133] zlibbioc_1.46.0 MASS_7.3-60
## [135] plyr_1.8.8 parallel_4.3.0
## [137] listenv_0.9.0 ggrepel_0.9.3
## [139] Biostrings_2.68.1 splines_4.3.0
## [141] hms_1.1.3 circlize_0.4.15
## [143] locfit_1.5-9.7 reshape2_1.4.4
## [145] ScaledMatrix_1.8.1 BiocVersion_3.17.1
## [147] XML_3.99-0.14 evaluate_0.21
## [149] BiocManager_1.30.20 nloptr_2.0.3
## [151] foreach_1.5.2 httpuv_1.6.11
## [153] tidyr_1.3.0 purrr_1.0.1
## [155] future_1.32.0 clue_0.3-64
## [157] scattermore_1.1 ashr_2.2-54
## [159] rsvd_1.0.5 broom_1.0.4
## [161] xtable_1.8-4 later_1.3.1
## [163] viridisLite_0.4.2 truncnorm_1.0-9
## [165] tibble_3.2.1 lmerTest_3.1-3
## [167] glmmTMB_1.1.7 memoise_2.0.1
## [169] beeswarm_0.4.0 AnnotationDbi_1.62.1
## [171] cluster_2.1.4 globals_0.16.2
## [173] GSEABase_1.62.0