Large-scale single-cell RNA-seq data manipulation with GDS files
Dr. Xiuwen Zheng (Genomics Research Center, AbbVie)
Jan 2023
~
~
Introduction
The SCArray package provides large-scale single-cell RNA-seq data manipulation using Genomic Data Structure (GDS) files. It combines dense/sparse matrices stored in GDS files and the Bioconductor infrastructure framework (SingleCellExperiment and DelayedArray) to provide out-of-memory data storage and manipulation using the R programming language. As shown in the figure, SCArray provides a SingleCellExperiment object for downstream data analyses. GDS is an alternative to HDF5. Unlike HDF5, GDS supports the direct storage of a sparse matrix without converting it to multiple vectors.
Figure 1: Workflow of SCArray
~
~
Installation
Requires R (>= v3.5.0), gdsfmt (>= v1.35.4)
Bioconductor repository
To install this package, start R and enter:
if (!requireNamespace("BiocManager", quietly=TRUE))
install.packages("BiocManager")
BiocManager::install("SCArray")~
~
Format conversion
Conversion from SingleCellExperiment
The SCArray package can convert a single-cell experiment object (SingleCellExperiment) to a GDS file using the function scConvGDS(). For example,
suppressPackageStartupMessages(library(SCArray))
suppressPackageStartupMessages(library(SingleCellExperiment))
# load a SingleCellExperiment object
fn <- system.file("extdata", "example.rds", package="SCArray")
sce <- readRDS(fn)
# convert to a GDS file
scConvGDS(sce, "test.gds")## Output: test.gds
## Compression: LZMA_RA
## Dimension: 250 x 500
## Assay List [counts]:
## counts !==================================================
## |+ counts { SparseReal32 250x500 LZMA_ra(13.2%), 77.2K }
## rowData:
## colData:
## Cell_ID [character]
## Cell_type [character]
## Timepoint [character]
## Done.## Object of class "SCArrayFileClass"
## File: test.gds (82.6K)
## + [ ] *
## |--+ feature.id { Str8 250 LZMA_ra(71.2%), 1.1K }
## |--+ sample.id { Str8 500 LZMA_ra(15.1%), 1.2K }
## |--+ counts { SparseReal32 250x500 LZMA_ra(13.2%), 77.2K }
## |--+ feature.data [ ]
## |--+ sample.data [ ]
## | |--+ Cell_ID { Str8 500 LZMA_ra(15.1%), 1.2K }
## | |--+ Cell_type { Str8 500 LZMA_ra(4.49%), 141B }
## | \--+ Timepoint { Str8 500 LZMA_ra(5.31%), 193B }
## \--+ meta.data [ ]Conversion from a matrix
The input of scConvGDS() can be a dense or sparse matrix for count data:
library(Matrix)
cnt <- matrix(0, nrow=4, ncol=8)
set.seed(100); cnt[sample.int(length(cnt), 8)] <- rpois(8, 4)
(cnt <- as(cnt, "sparseMatrix"))## 4 x 8 sparse Matrix of class "dgCMatrix"
##
## [1,] . . . . . . . 6
## [2,] 3 1 . . . 4 . .
## [3,] . . . . . 3 . 4
## [4,] 4 . 3 . . . . .## Warning in scConvGDS(cnt, "test.gds"): rownames=NULL, use c("g1", "g2", ...) instead.## Warning in scConvGDS(cnt, "test.gds"): colnames=NULL, use c("c1", "c2", ...) instead.## Output: test.gds
## Compression: LZMA_RA
## Dimension: 4 x 8
## Assay List [counts]:
## counts !==================================================
## |+ counts { SparseReal32 4x8 LZMA_ra(159.4%), 109B }
## Done.~
~
Examples
When a single-cell GDS file is available, users can use scExperiment() to load a SingleCellExperiment object from the GDS file. The assay data in the SingleCellExperiment object are DelayedMatrix objects.
# a GDS file in the SCArray package
(fn <- system.file("extdata", "example.gds", package="SCArray"))## [1] "/tmp/Rtmpx27uIm/Rinst38931b643b71fe/SCArray/extdata/example.gds"## class: SingleCellExperiment
## dim: 1000 850
## metadata(0):
## assays(1): counts
## rownames(1000): MRPL20 GNB1 ... RPS4Y1 CD24
## rowData names(0):
## colnames(850): 1772122_301_C02 1772122_180_E05 ... 1772122_180_B06 1772122_180_D09
## colData names(3): Cell_ID Cell_type Timepoint
## reducedDimNames(0):
## mainExpName: NULL
## altExpNames(0):## <1000 x 850> sparse SC_GDSMatrix object of type "double":
## 1772122_301_C02 1772122_180_E05 1772122_300_H02 ... 1772122_180_B06 1772122_180_D09
## MRPL20 3 2 3 . 0 2
## GNB1 11 6 15 . 0 0
## RPL22 3 5 7 . 6 6
## PARK7 1 7 3 . 2 2
## ENO1 8 19 20 . 7 4
## ... . . . . . .
## SSR4 0 6 3 . 5 1
## RPL10 11 4 8 . 1 3
## SLC25A6-loc1 4 5 5 . 3 1
## RPS4Y1 0 5 0 . 2 4
## CD24 18 3 7 . 0 2## DataFrame with 850 rows and 3 columns
## Cell_ID Cell_type Timepoint
## <character> <character> <character>
## 1772122_301_C02 1772122_301_C02 eNb1 day_35
## 1772122_180_E05 1772122_180_E05 eNb1 day_35
## 1772122_300_H02 1772122_300_H02 eNb1 day_35
## 1772122_180_B09 1772122_180_B09 eNb1 day_35
## 1772122_180_G04 1772122_180_G04 eNb1 day_35
## ... ... ... ...
## 1772122_181_F11 1772122_181_F11 eRgld day_35
## 1772122_181_E02 1772122_181_E02 eRgld day_35
## 1772122_180_C03 1772122_180_C03 eRgld day_35
## 1772122_180_B06 1772122_180_B06 eRgld day_35
## 1772122_180_D09 1772122_180_D09 eRgld day_35## DataFrame with 1000 rows and 0 columns~
~
Data Manipulation and Analysis
SCArray provides a SingleCellExperiment object for downstream data analyses. At first, we create a log count matrix logcnt from the count matrix. Note that logcnt is also a DelayedMatrix without actually generating the whole matrix.
## <1000 x 850> sparse SC_GDSMatrix object of type "double":
## 1772122_301_C02 1772122_180_E05 1772122_300_H02 ... 1772122_180_B06 1772122_180_D09
## MRPL20 2.000000 1.584963 2.000000 . 0.000000 1.584963
## GNB1 3.584963 2.807355 4.000000 . 0.000000 0.000000
## RPL22 2.000000 2.584963 3.000000 . 2.807355 2.807355
## PARK7 1.000000 3.000000 2.000000 . 1.584963 1.584963
## ENO1 3.169925 4.321928 4.392317 . 3.000000 2.321928
## ... . . . . . .
## SSR4 0.000000 2.807355 2.000000 . 2.584963 1.000000
## RPL10 3.584963 2.321928 3.169925 . 1.000000 2.000000
## SLC25A6-loc1 2.321928 2.584963 2.584963 . 2.000000 1.000000
## RPS4Y1 0.000000 2.584963 0.000000 . 1.584963 2.321928
## CD24 4.247928 2.000000 3.000000 . 0.000000 1.584963Formally, we call logNormCounts() in the scuttle package to normalize the raw counts.
## <1000 x 850> sparse SC_GDSMatrix object of type "double":
## 1772122_301_C02 1772122_180_E05 1772122_300_H02 ... 1772122_180_B06 1772122_180_D09
## MRPL20 2.318287 1.425944 1.693629 . 0.000000 1.835269
## GNB1 3.965364 2.599486 3.605669 . 0.000000 0.000000
## RPL22 2.318287 2.383291 2.635589 . 3.463948 3.121867
## PARK7 1.219792 2.787466 1.693629 . 2.119300 1.835269
## ENO1 3.540158 4.089631 3.990770 . 3.667541 2.617477
## ... . . . . . .
## SSR4 0.000000 2.599486 1.693629 . 3.226828 1.191683
## RPL10 3.965364 2.128889 2.798918 . 1.418144 2.278749
## SLC25A6-loc1 2.659132 2.383291 2.240159 . 2.589103 1.191683
## RPS4Y1 0.000000 2.383291 0.000000 . 2.119300 2.617477
## CD24 4.639485 1.819813 2.635589 . 0.000000 1.8352691. Row and Column Summarization
The DelayedMatrixStats package provides summarization functions operating on rows and columns of DelayedMatrix objects. SCArray has provided the optimized implementations for the row and column summarization. For example, we can calculate the mean for each column or row of the log count matrix.
## Named num [1:850] 1.51 1.95 2.25 1.95 1.75 ...
## - attr(*, "names")= chr [1:850] "1772122_301_C02" "1772122_180_E05" "1772122_300_H02" "1772122_180_B09" ...## Named num [1:1000] 1.27 1.51 2.62 1.98 3.75 ...
## - attr(*, "names")= chr [1:1000] "MRPL20" "GNB1" "RPL22" "PARK7" ...## mean var
## [1,] 1.271806 0.9867109
## [2,] 1.514454 1.0162891
## [3,] 2.623583 1.0823599
## [4,] 1.976739 1.1728957
## [5,] 3.754426 1.0599877
## [6,] 1.315402 0.92870272. PCA analysis
The scater package can perform the Principal component analysis (PCA) on the normalized cell data.
scater::runPCA() will call the function beachmat::realizeFileBackedMatrix() internally to realize a scaled and centered DelayedMatrix into its corresponding in-memory format, so it is memory-intensive for large-scale PCA.
Instead, the SCArray package provides scRunPCA() for reducing the memory usage in large-scale PCA by perform SVD on the relatively small covariance matrix.
## Select top 500 features with the highest variances ...
## Start PCA on the covariance matrix ...plotReducedDim() plots cell-level reduced dimension results (PCA) stored in the SingleCellExperiment object:
3. UMAP analysis
The scater package can perform the uniform manifold approximation and projection (UMAP) for the cell data, based on the data in a SingleCellExperiment object.
plotReducedDim() plots cell-level reduced dimension results (UMAP) stored in the SingleCellExperiment object:
~
~
Miscellaneous
Debugging
options(SCArray.verbose=TRUE) is used to enable displaying debug information when calling the functions in the SCArray packages. For example,
## Calling SCArray:::x_rowMeans() with SC_GDSMatrix [1000x850] ...## Named num [1:1000] 1.27 1.51 2.62 1.98 3.75 ...
## - attr(*, "names")= chr [1:1000] "MRPL20" "GNB1" "RPL22" "PARK7" ...~
~
Session Information
## R version 4.3.1 (2023-06-16)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 22.04.2 LTS
##
## Matrix products: default
## BLAS: /home/biocbuild/bbs-3.17-bioc/R/lib/libRblas.so
## LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.10.0
##
## locale:
## [1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C LC_TIME=en_GB
## [4] LC_COLLATE=C LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## [7] LC_PAPER=en_US.UTF-8 LC_NAME=C LC_ADDRESS=C
## [10] LC_TELEPHONE=C LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## time zone: America/New_York
## tzcode source: system (glibc)
##
## attached base packages:
## [1] stats4 stats graphics grDevices utils datasets methods base
##
## other attached packages:
## [1] scater_1.28.0 ggplot2_3.4.2 scuttle_1.10.2
## [4] SingleCellExperiment_1.22.0 SummarizedExperiment_1.30.2 Biobase_2.60.0
## [7] GenomicRanges_1.52.0 GenomeInfoDb_1.36.1 SCArray_1.8.4
## [10] DelayedArray_0.26.7 S4Arrays_1.0.5 abind_1.4-5
## [13] IRanges_2.34.1 S4Vectors_0.38.1 MatrixGenerics_1.12.3
## [16] matrixStats_1.0.0 BiocGenerics_0.46.0 Matrix_1.6-0
## [19] gdsfmt_1.36.1
##
## loaded via a namespace (and not attached):
## [1] tidyselect_1.2.0 viridisLite_0.4.2 dplyr_1.1.2 vipor_0.4.5
## [5] farver_2.1.1 viridis_0.6.4 bitops_1.0-7 fastmap_1.1.1
## [9] RCurl_1.98-1.12 digest_0.6.33 rsvd_1.0.5 lifecycle_1.0.3
## [13] magrittr_2.0.3 compiler_4.3.1 rlang_1.1.1 sass_0.4.7
## [17] tools_4.3.1 utf8_1.2.3 yaml_2.3.7 knitr_1.43
## [21] FNN_1.1.3.2 labeling_0.4.2 BiocParallel_1.34.2 withr_2.5.0
## [25] grid_4.3.1 fansi_1.0.4 beachmat_2.16.0 colorspace_2.1-0
## [29] scales_1.2.1 cli_3.6.1 rmarkdown_2.23 crayon_1.5.2
## [33] generics_0.1.3 DelayedMatrixStats_1.22.5 ggbeeswarm_0.7.2 cachem_1.0.8
## [37] zlibbioc_1.46.0 parallel_4.3.1 XVector_0.40.0 vctrs_0.6.3
## [41] jsonlite_1.8.7 BiocSingular_1.16.0 BiocNeighbors_1.18.0 ggrepel_0.9.3
## [45] irlba_2.3.5.1 beeswarm_0.4.0 jquerylib_0.1.4 glue_1.6.2
## [49] codetools_0.2-19 cowplot_1.1.1 uwot_0.1.16 gtable_0.3.3
## [53] ScaledMatrix_1.8.1 munsell_0.5.0 tibble_3.2.1 pillar_1.9.0
## [57] htmltools_0.5.6 GenomeInfoDbData_1.2.10 R6_2.5.1 sparseMatrixStats_1.12.2
## [61] evaluate_0.21 lattice_0.21-8 highr_0.10 bslib_0.5.1
## [65] Rcpp_1.0.11 gridExtra_2.3 xfun_0.40 pkgconfig_2.0.3