Fit Gamma-Poisson Generalized Linear Models Reliably.
The core design aims of gmlGamPoi
are:
DESeq2
or edgeR
You can install the release version of glmGamPoi from BioConductor:
if (!requireNamespace("BiocManager", quietly = TRUE))
install.packages("BiocManager")
BiocManager::install("glmGamPoi")
For the latest developments, see the GitHub repo.
Load the glmGamPoi package
library(glmGamPoi)
To fit a single Gamma-Poisson GLM do:
# overdispersion = 1/size
counts <- rnbinom(n = 10, mu = 5, size = 1/0.7)
# design = ~ 1 means that an intercept-only model is fit
fit <- glm_gp(counts, design = ~ 1)
fit
#> glmGamPoiFit object:
#> The data had 1 rows and 10 columns.
#> A model with 1 coefficient was fitted.
# Internally fit is just a list:
as.list(fit)[1:2]
#> $Beta
#> Intercept
#> [1,] 1.504077
#>
#> $overdispersions
#> [1] 0.3792855
The glm_gp()
function returns a list with the results of the fit. Most importantly, it contains the estimates for the coefficients β and the overdispersion.
Fitting repeated Gamma-Poisson GLMs for each gene of a single cell dataset is just as easy:
I will first load an example dataset using the TENxPBMCData
package. The dataset has 33,000 genes and 4340 cells. It takes roughly 1.5 minutes to fit the Gamma-Poisson model on the full dataset. For demonstration purposes, I will subset the dataset to 300 genes, but keep the 4340 cells:
library(SummarizedExperiment)
library(DelayedMatrixStats)
# The full dataset with 33,000 genes and 4340 cells
# The first time this is run, it will download the data
pbmcs <- TENxPBMCData::TENxPBMCData("pbmc4k")
#> snapshotDate(): 2020-10-02
#> see ?TENxPBMCData and browseVignettes('TENxPBMCData') for documentation
#> loading from cache
# I want genes where at least some counts are non-zero
non_empty_rows <- which(rowSums2(assay(pbmcs)) > 0)
pbmcs_subset <- pbmcs[sample(non_empty_rows, 300), ]
pbmcs_subset
#> class: SingleCellExperiment
#> dim: 300 4340
#> metadata(0):
#> assays(1): counts
#> rownames(300): ENSG00000126457 ENSG00000109832 ... ENSG00000143819
#> ENSG00000188243
#> rowData names(3): ENSEMBL_ID Symbol_TENx Symbol
#> colnames: NULL
#> colData names(11): Sample Barcode ... Individual Date_published
#> reducedDimNames(0):
#> altExpNames(0):
I call glm_gp()
to fit one GLM model for each gene and force the calculation to happen in memory.
fit <- glm_gp(pbmcs_subset, on_disk = FALSE)
summary(fit)
#> glmGamPoiFit object:
#> The data had 300 rows and 4340 columns.
#> A model with 1 coefficient was fitted.
#> The design formula is: Y~1
#>
#> Beta:
#> Min 1st Qu. Median 3rd Qu. Max
#> Intercept -8.51 -6.57 -3.91 -2.59 0.903
#>
#> deviance:
#> Min 1st Qu. Median 3rd Qu. Max
#> 14 86.8 657 1686 5507
#>
#> overdispersion:
#> Min 1st Qu. Median 3rd Qu. Max
#> 0 1.65e-13 0.288 1.84 24687
#>
#> Shrunken quasi-likelihood overdispersion:
#> Min 1st Qu. Median 3rd Qu. Max
#> 0.707 0.991 1 1.04 7.45
#>
#> size_factors:
#> Min 1st Qu. Median 3rd Qu. Max
#> 0.117 0.738 1.01 1.32 14.5
#>
#> Mu:
#> Min 1st Qu. Median 3rd Qu. Max
#> 2.34e-05 0.00142 0.0185 0.0779 35.8
I compare my method (in-memory and on-disk) with DESeq2 and edgeR. Both are classical methods for analyzing RNA-Seq datasets and have been around for almost 10 years. Note that both tools can do a lot more than just fitting the Gamma-Poisson model, so this benchmark only serves to give a general impression of the performance.
# Explicitly realize count matrix in memory so that it is a fair comparison
pbmcs_subset <- as.matrix(assay(pbmcs_subset))
model_matrix <- matrix(1, nrow = ncol(pbmcs_subset))
bench::mark(
glmGamPoi_in_memory = {
glm_gp(pbmcs_subset, design = model_matrix, on_disk = FALSE)
}, glmGamPoi_on_disk = {
glm_gp(pbmcs_subset, design = model_matrix, on_disk = TRUE)
}, DESeq2 = suppressMessages({
dds <- DESeq2::DESeqDataSetFromMatrix(pbmcs_subset,
colData = data.frame(name = seq_len(4340)),
design = ~ 1)
dds <- DESeq2::estimateSizeFactors(dds, "poscounts")
dds <- DESeq2::estimateDispersions(dds, quiet = TRUE)
dds <- DESeq2::nbinomWaldTest(dds, minmu = 1e-6)
}), edgeR = {
edgeR_data <- edgeR::DGEList(pbmcs_subset)
edgeR_data <- edgeR::calcNormFactors(edgeR_data)
edgeR_data <- edgeR::estimateDisp(edgeR_data, model_matrix)
edgeR_fit <- edgeR::glmFit(edgeR_data, design = model_matrix)
}, check = FALSE, min_iterations = 3
)
#> # A tibble: 4 x 6
#> expression min median `itr/sec` mem_alloc `gc/sec`
#> <bch:expr> <bch:tm> <bch:tm> <dbl> <bch:byt> <dbl>
#> 1 glmGamPoi_in_memory 2.15s 2.36s 0.434 NA 1.45
#> 2 glmGamPoi_on_disk 6.72s 6.9s 0.145 NA 0.726
#> 3 DESeq2 29.23s 29.49s 0.0334 NA 0.312
#> 4 edgeR 11.8s 12.94s 0.0789 NA 0.789
On this dataset, glmGamPoi
is more than 5 times faster than edgeR
and more than 18 times faster than DESeq2
. glmGamPoi
does not use approximations to achieve this performance increase. The performance comes from an optimized algorithm for inferring the overdispersion for each gene. It is tuned for datasets typically encountered in single RNA-seq with many samples and many small counts, by avoiding duplicate calculations.
To demonstrate that the method does not sacrifice accuracy, I compare the parameters that each method estimates. The means and β coefficients are identical, but that the overdispersion estimates from glmGamPoi
are more reliable:
# Results with my method
fit <- glm_gp(pbmcs_subset, design = model_matrix, on_disk = FALSE)
# DESeq2
dds <- DESeq2::DESeqDataSetFromMatrix(pbmcs_subset,
colData = data.frame(name = seq_len(4340)),
design = ~ 1)
sizeFactors(dds) <- fit$size_factors
dds <- DESeq2::estimateDispersions(dds, quiet = TRUE)
dds <- DESeq2::nbinomWaldTest(dds, minmu = 1e-6)
#edgeR
edgeR_data <- edgeR::DGEList(pbmcs_subset, lib.size = fit$size_factors)
edgeR_data <- edgeR::estimateDisp(edgeR_data, model_matrix)
edgeR_fit <- edgeR::glmFit(edgeR_data, design = model_matrix)
I am comparing the gene-wise estimates of the coefficients from all three methods. Points on the diagonal line are identical. The inferred Beta coefficients and gene means agree well between the methods, however the overdispersion differs quite a bit. DESeq2
has problems estimating most of the overdispersions and sets them to 1e-8
. edgeR
only approximates the overdispersions which explains the variation around the overdispersions calculated with glmGamPoi
.
The method scales linearly, with the number of rows and columns in the dataset. For example: fitting the full pbmc4k
dataset with subsampling on a modern MacBook Pro in-memory takes ~1 minute and on-disk a little over 4 minutes. Fitting the pbmc68k
(17x the size) takes ~73 minutes (17x the time) on-disk.
glmGamPoi
provides an interface to do quasi-likelihood ratio testing to identify differentially expressed genes:
# Create random categorical assignment to demonstrate DE
group <- sample(c("Group1", "Group2"), size = ncol(pbmcs_subset), replace = TRUE)
# Fit model with group vector as design
fit <- glm_gp(pbmcs_subset, design = group)
# Compare against model without group
res <- test_de(fit, reduced_design = ~ 1)
# Look at first 6 genes
head(res)
#> name pval adj_pval f_statistic df1 df2 lfc
#> 1 ENSG00000126457 0.2385897 0.8863222 1.389282668 1 4420.177 NA
#> 2 ENSG00000109832 0.6491580 0.9275674 0.206991593 1 4420.177 NA
#> 3 ENSG00000237339 0.4375426 0.9053288 0.602828037 1 4420.177 NA
#> 4 ENSG00000075234 0.3118470 0.8877735 1.023070967 1 4420.177 NA
#> 5 ENSG00000161057 0.9429562 0.9870834 0.005120673 1 4420.177 NA
#> 6 ENSG00000151366 0.5245737 0.9225492 0.404956209 1 4420.177 NA
The p-values agree well with the ones that edgeR
is calculating. This is because glmGamPoi
uses the same framework of quasi-likelihood ratio tests that was invented by edgeR
and is described in Lund et al. (2012).
model_matrix <- model.matrix(~ group, data = data.frame(group = group))
edgeR_data <- edgeR::DGEList(pbmcs_subset)
edgeR_data <- edgeR::calcNormFactors(edgeR_data)
edgeR_data <- edgeR::estimateDisp(edgeR_data, design = model_matrix)
edgeR_fit <- edgeR::glmQLFit(edgeR_data, design = model_matrix)
edgeR_test <- edgeR::glmQLFTest(edgeR_fit, coef = 2)
edgeR_res <- edgeR::topTags(edgeR_test, sort.by = "none", n = nrow(pbmcs_subset))
Be very careful how you interpret the p-values of a single cell experiment. Cells that come from one individual are not independent replicates. That means that you cannot turn your RNA-seq experiment with 3 treated and 3 control samples into a 3000 vs 3000 experiment by measuring 1000 cells per sample. The actual unit of replication are still the 3 samples in each condition.
Nonetheless, single cell data is valuable because it allows you to compare the effect of a treatment on specific cell types. The simplest way to do such a test is called pseudobulk. This means that the data is subset to the cells of a specific cell type. Then the counts of cells from the same sample are combined to form a “pseudobulk” sample. The test_de()
function of glmGamPoi supports this feature directly through the pseudobulk_by
and subset_to
parameters:
# say we have cell type labels for each cell and know from which sample they come originally
sample_labels <- rep(paste0("sample_", 1:6), length = ncol(pbmcs_subset))
cell_type_labels <- sample(c("T-cells", "B-cells", "Macrophages"), ncol(pbmcs_subset), replace = TRUE)
test_de(fit, contrast = Group1 - Group2,
pseudobulk_by = sample_labels,
subset_to = cell_type_labels == "T-cells",
n_max = 4, sort_by = pval, decreasing = FALSE)
#> name pval adj_pval f_statistic df1 df2 lfc
#> 218 ENSG00000158411 0.03539352 1 5.609945 1 12.0646 -15.89212
#> 96 ENSG00000105610 0.04160948 1 5.195969 1 12.0646 1156.23142
#> 110 ENSG00000134539 0.04802063 1 4.840090 1 12.0646 10.43825
#> 107 ENSG00000162642 0.06805301 1 4.015867 1 12.0646 -16.94800
sessionInfo()
#> R version 4.0.3 (2020-10-10)
#> Platform: x86_64-pc-linux-gnu (64-bit)
#> Running under: Ubuntu 18.04.5 LTS
#>
#> Matrix products: default
#> BLAS: /home/biocbuild/bbs-3.12-bioc/R/lib/libRblas.so
#> LAPACK: /home/biocbuild/bbs-3.12-bioc/R/lib/libRlapack.so
#>
#> locale:
#> [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
#> [3] LC_TIME=en_US.UTF-8 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
#>
#> attached base packages:
#> [1] parallel stats4 stats graphics grDevices utils datasets
#> [8] methods base
#>
#> other attached packages:
#> [1] TENxPBMCData_1.7.0 HDF5Array_1.18.0
#> [3] rhdf5_2.34.0 SingleCellExperiment_1.12.0
#> [5] DelayedMatrixStats_1.12.0 DelayedArray_0.16.0
#> [7] Matrix_1.2-18 SummarizedExperiment_1.20.0
#> [9] Biobase_2.50.0 GenomicRanges_1.42.0
#> [11] GenomeInfoDb_1.26.0 IRanges_2.24.0
#> [13] S4Vectors_0.28.0 BiocGenerics_0.36.0
#> [15] MatrixGenerics_1.2.0 matrixStats_0.57.0
#> [17] glmGamPoi_1.2.0 BiocStyle_2.18.0
#>
#> loaded via a namespace (and not attached):
#> [1] bitops_1.0-6 bit64_4.0.5
#> [3] RColorBrewer_1.1-2 httr_1.4.2
#> [5] tools_4.0.3 utf8_1.1.4
#> [7] R6_2.4.1 colorspace_1.4-1
#> [9] DBI_1.1.0 rhdf5filters_1.2.0
#> [11] tidyselect_1.1.0 DESeq2_1.30.0
#> [13] bit_4.0.4 curl_4.3
#> [15] compiler_4.0.3 cli_2.1.0
#> [17] bookdown_0.21 scales_1.1.1
#> [19] bench_1.1.1 genefilter_1.72.0
#> [21] rappdirs_0.3.1 stringr_1.4.0
#> [23] digest_0.6.27 rmarkdown_2.5
#> [25] XVector_0.30.0 pkgconfig_2.0.3
#> [27] htmltools_0.5.0 sparseMatrixStats_1.2.0
#> [29] limma_3.46.0 dbplyr_1.4.4
#> [31] fastmap_1.0.1 rlang_0.4.8
#> [33] RSQLite_2.2.1 shiny_1.5.0
#> [35] generics_0.0.2 BiocParallel_1.24.0
#> [37] dplyr_1.0.2 RCurl_1.98-1.2
#> [39] magrittr_1.5 GenomeInfoDbData_1.2.4
#> [41] fansi_0.4.1 Rcpp_1.0.5
#> [43] munsell_0.5.0 Rhdf5lib_1.12.0
#> [45] lifecycle_0.2.0 edgeR_3.32.0
#> [47] stringi_1.5.3 yaml_2.2.1
#> [49] zlibbioc_1.36.0 BiocFileCache_1.14.0
#> [51] AnnotationHub_2.22.0 grid_4.0.3
#> [53] blob_1.2.1 promises_1.1.1
#> [55] ExperimentHub_1.16.0 crayon_1.3.4
#> [57] lattice_0.20-41 beachmat_2.6.0
#> [59] splines_4.0.3 annotate_1.68.0
#> [61] magick_2.5.0 locfit_1.5-9.4
#> [63] knitr_1.30 pillar_1.4.6
#> [65] geneplotter_1.68.0 XML_3.99-0.5
#> [67] glue_1.4.2 BiocVersion_3.12.0
#> [69] evaluate_0.14 BiocManager_1.30.10
#> [71] vctrs_0.3.4 httpuv_1.5.4
#> [73] gtable_0.3.0 purrr_0.3.4
#> [75] assertthat_0.2.1 ggplot2_3.3.2
#> [77] xfun_0.18 mime_0.9
#> [79] xtable_1.8-4 later_1.1.0.1
#> [81] survival_3.2-7 tibble_3.0.4
#> [83] AnnotationDbi_1.52.0 memoise_1.1.0
#> [85] ellipsis_0.3.1 interactiveDisplayBase_1.28.0