Last updated: 2022-09-14

Checks: 7 0

Knit directory: chromap_vs_cellranger_scATAC_exploration_10x/

This reproducible R Markdown analysis was created with workflowr (version 1.7.0). The Checks tab describes the reproducibility checks that were applied when the results were created. The Past versions tab lists the development history.

R Markdown file: up-to-date

Great! Since the R Markdown file has been committed to the Git repository, you know the exact version of the code that produced these results.

Environment: empty

Great job! The global environment was empty. Objects defined in the global environment can affect the analysis in your R Markdown file in unknown ways. For reproduciblity it’s best to always run the code in an empty environment.

Seed: set.seed(20220912)

The command set.seed(20220912) was run prior to running the code in the R Markdown file. Setting a seed ensures that any results that rely on randomness, e.g. subsampling or permutations, are reproducible.

Session information: recorded

Great job! Recording the operating system, R version, and package versions is critical for reproducibility.

Cache: none

Nice! There were no cached chunks for this analysis, so you can be confident that you successfully produced the results during this run.

File paths: relative

Great job! Using relative paths to the files within your workflowr project makes it easier to run your code on other machines.

Repository version: 10fdcb0

Great! You are using Git for version control. Tracking code development and connecting the code version to the results is critical for reproducibility.

The results in this page were generated with repository version 10fdcb0. See the Past versions tab to see a history of the changes made to the R Markdown and HTML files.

Note that you need to be careful to ensure that all relevant files for the analysis have been committed to Git prior to generating the results (you can use wflow_publish or wflow_git_commit). workflowr only checks the R Markdown file, but you know if there are other scripts or data files that it depends on. Below is the status of the Git repository when the results were generated: 


working directory clean

Note that any generated files, e.g. HTML, png, CSS, etc., are not included in this status report because it is ok for generated content to have uncommitted changes.

These are the previous versions of the repository in which changes were made to the R Markdown (analysis/CellrangerSignacIntegration.Rmd) and HTML (docs/CellrangerSignacIntegration.html) files. If you’ve configured a remote Git repository (see ?wflow_git_remote), click on the hyperlinks in the table below to view the files as they were in that past version.

File Version Author Date Message
Rmd 10fdcb0 jeremymsimon 2022-09-14 Initial commit

Load packages and saved workspace

library(tidyverse)
── Attaching packages ─────────────────────────────────────── tidyverse 1.3.1 ──
✔ ggplot2 3.3.6     ✔ purrr   0.3.4
✔ tibble  3.1.8     ✔ dplyr   1.0.9
✔ tidyr   1.2.0     ✔ stringr 1.4.0
✔ readr   2.1.2     ✔ forcats 0.5.1
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
library(GenomicRanges)
Loading required package: stats4
Loading required package: BiocGenerics

Attaching package: 'BiocGenerics'
The following objects are masked from 'package:dplyr':

    combine, intersect, setdiff, union
The following objects are masked from 'package:stats':

    IQR, mad, sd, var, xtabs
The following objects are masked from 'package:base':

    anyDuplicated, append, as.data.frame, basename, cbind, colnames,
    dirname, do.call, duplicated, eval, evalq, Filter, Find, get, grep,
    grepl, intersect, is.unsorted, lapply, Map, mapply, match, mget,
    order, paste, pmax, pmax.int, pmin, pmin.int, Position, rank,
    rbind, Reduce, rownames, sapply, setdiff, sort, table, tapply,
    union, unique, unsplit, which.max, which.min
Loading required package: S4Vectors

Attaching package: 'S4Vectors'
The following objects are masked from 'package:dplyr':

    first, rename
The following object is masked from 'package:tidyr':

    expand
The following objects are masked from 'package:base':

    expand.grid, I, unname
Loading required package: IRanges

Attaching package: 'IRanges'
The following objects are masked from 'package:dplyr':

    collapse, desc, slice
The following object is masked from 'package:purrr':

    reduce
Loading required package: GenomeInfoDb
library(Seurat)
Attaching SeuratObject
library(Signac)
library(EnsDb.Hsapiens.v86)
Loading required package: ensembldb
Loading required package: GenomicFeatures
Loading required package: AnnotationDbi
Loading required package: Biobase
Welcome to Bioconductor

    Vignettes contain introductory material; view with
    'browseVignettes()'. To cite Bioconductor, see
    'citation("Biobase")', and for packages 'citation("pkgname")'.

Attaching package: 'AnnotationDbi'
The following object is masked from 'package:dplyr':

    select
Loading required package: AnnotationFilter

Attaching package: 'ensembldb'
The following object is masked from 'package:dplyr':

    filter
The following object is masked from 'package:stats':

    filter
load("Cellranger_HGMM_PBMC_seurat_090222_QC.RData")

Create union set of peaks

This in my opinion is the source of a major shortcoming of cellranger, and results in a lot of redundant/circular logic and processing. If the question of “what constitutes a cell?” depends on how much signal and enrichment we get over genomic features, then cellranger has already decided which minimally-viable cells to retain before we’ve even quantified the signal over the final set of features. In other words, the accepted workflow calls peaks on each sample, quantifies and filters out some cells based on those peaks, and is completely unaware of information from other cells/samples/replicates/conditions run separately. We therefore need to do most of this all over again here; we define a union set and run FeatureMatrix to re-do all of our feature counts per cell.

In my opinion, it would be far better to identify all peaks based on all samples and experimental conditions, quantify the number of fragments for each cell barcode under those features once, then filter for QC and proceed with the analysis.

Both approaches share the additional limitation that some regulatory elements may only be present or utilized in a small fraction of cells, or otherwise are cell-type- or condition-specific. Calling peaks essentially in a pseudo-bulk fashion like this may lack the power to detect sites that aren’t largely ubiquitous. This is an area I believe we in the field can and should improve upon.

Import cellranger peak calls

peaks.HGMM <- read.table(
  file = "10x_HGMM_cellranger/outs/peaks.bed",
  col.names = c("chr", "start", "end")
)
peaks.PBMC <- read.table(
  file = "10x_PBMC_cellranger/outs/peaks.bed",
  col.names = c("chr", "start", "end")
)

Convert to genomic ranges

gr.HGMM <- makeGRangesFromDataFrame(peaks.HGMM)
gr.PBMC <- makeGRangesFromDataFrame(peaks.PBMC)

Import cCREs from Zhang. et al 2021 scATAC atlas study

p <- as.data.frame(read.table("cCRE_hg38.bed",header=F,sep="\t"))
colnames(p) <- c("chr","start","stop")
cre <- makeGRangesFromDataFrame(p)

Remove non standard chromosomes

gr.HGMM <- keepStandardChromosomes(gr.HGMM,pruning.mode="coarse")
gr.PBMC <- keepStandardChromosomes(gr.PBMC,pruning.mode="coarse")
gr.cre <- keepStandardChromosomes(cre,pruning.mode="coarse")

Create a unified set of peaks to quantify in each dataset

combined.peaks <- reduce(x = c(gr.HGMM, gr.PBMC, gr.cre))

# Filter out bad peaks based on length
peakwidths <- width(combined.peaks)
combined.peaks <- combined.peaks[peakwidths  < 10000 & peakwidths > 20]

Compute feature matrix for combined peaks

HGMM.counts <- FeatureMatrix(
  fragments = Fragments(HGMM_cr_seurat),
  features = combined.peaks,
  cells = colnames(HGMM_cr_seurat)
)
Extracting reads overlapping genomic regions
PBMC.counts <- FeatureMatrix(
  fragments = Fragments(PBMC_cr_seurat),
  features = combined.peaks,
  cells = colnames(PBMC_cr_seurat)
)
Extracting reads overlapping genomic regions

Create new chromatin assays and seurat objects

HGMM_assay <- CreateChromatinAssay(HGMM.counts, fragments = Fragments(HGMM_cr_seurat))
HGMM_seurat <- CreateSeuratObject(HGMM_assay, assay = "ATAC", meta.data=as.data.frame(HGMM_cr_seurat@meta.data))
Warning: Keys should be one or more alphanumeric characters followed by an
underscore, setting key from atac to atac_
HGMM_seurat$Sample <- "HGMM"

PBMC_assay <- CreateChromatinAssay(PBMC.counts, fragments = Fragments(PBMC_cr_seurat))
PBMC_seurat <- CreateSeuratObject(PBMC_assay, assay = "ATAC", meta.data=as.data.frame(PBMC_cr_seurat@meta.data))
Warning: Keys should be one or more alphanumeric characters followed by an
underscore, setting key from atac to atac_
PBMC_seurat$Sample <- "PBMC"

Merge seurat objects

cr.combined <- merge(HGMM_seurat, y = PBMC_seurat)

Subset for features observed in at least 20 cells

filtFeatures <- FindTopFeatures(cr.combined,min.cutoff=20)
cr.combined <- subset(cr.combined, features = VariableFeatures(filtFeatures))

Split by sample into a list

cr.list <- SplitObject(cr.combined, split.by="Sample")

Iterate over sample list, subset based on QC metrics, and generate lsi embeddings

for (i in 1:length(cr.list)) {
    cr.list[[i]] <- subset(x = cr.list[[i]],
        features = VariableFeatures(filtFeatures), 
        subset = nCount_ATAC > 1000 &
            nCount_ATAC < 100000 &
            FRiP > 0.15 &
            blacklist_fraction < 0.05 &
            nucleosome_signal < 4 &
            TSS.enrichment > 2
        )

    cr.list[[i]] <- FindTopFeatures(cr.list[[i]], min.cutoff = 'q0')
    cr.list[[i]] <- RunTFIDF(cr.list[[i]], assay="ATAC")
    cr.list[[i]] <- RunSVD(cr.list[[i]])
    cr.list[[i]] <- RunUMAP(cr.list[[i]], reduction = "lsi", dims = 2:30)
}
Performing TF-IDF normalization
Warning in RunTFIDF.default(object = GetAssayData(object = object, slot =
"counts"), : Some features contain 0 total counts
Running SVD
Scaling cell embeddings
Warning: The default method for RunUMAP has changed from calling Python UMAP via reticulate to the R-native UWOT using the cosine metric
To use Python UMAP via reticulate, set umap.method to 'umap-learn' and metric to 'correlation'
This message will be shown once per session
05:56:36 UMAP embedding parameters a = 0.9922 b = 1.112
05:56:36 Read 6639 rows and found 29 numeric columns
05:56:36 Using Annoy for neighbor search, n_neighbors = 30
05:56:36 Building Annoy index with metric = cosine, n_trees = 50
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
05:56:38 Writing NN index file to temp file /tmp/RtmpwGLNhI/file3b98f1355a0f1f
05:56:38 Searching Annoy index using 1 thread, search_k = 3000
05:56:41 Annoy recall = 100%
05:56:46 Commencing smooth kNN distance calibration using 1 thread
05:56:52 Initializing from normalized Laplacian + noise
05:56:53 Commencing optimization for 500 epochs, with 272968 positive edges
05:57:08 Optimization finished
Performing TF-IDF normalization
Warning in RunTFIDF.default(object = GetAssayData(object = object, slot =
"counts"), : Some features contain 0 total counts
Running SVD
Scaling cell embeddings
06:05:15 UMAP embedding parameters a = 0.9922 b = 1.112
06:05:15 Read 9939 rows and found 29 numeric columns
06:05:15 Using Annoy for neighbor search, n_neighbors = 30
06:05:15 Building Annoy index with metric = cosine, n_trees = 50
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
06:05:17 Writing NN index file to temp file /tmp/RtmpwGLNhI/file3b98f1f8a05eb
06:05:17 Searching Annoy index using 1 thread, search_k = 3000
06:05:22 Annoy recall = 100%
06:05:26 Commencing smooth kNN distance calibration using 1 thread
06:05:31 Initializing from normalized Laplacian + noise
06:05:32 Commencing optimization for 500 epochs, with 405586 positive edges
06:05:49 Optimization finished

Subset and process the combined dataset

cr.combined <- subset(
  x = cr.combined,
  features = VariableFeatures(filtFeatures), 
  subset = nCount_ATAC > 1000 &
    nCount_ATAC < 100000 &
    FRiP > 0.15 &
    blacklist_fraction < 0.05 &
    nucleosome_signal < 4 &
    TSS.enrichment > 2
)

Get filtered dimensions

table(cr.combined$Sample)

HGMM PBMC 
6639 9939 
lapply(cr.list, dim)
$HGMM
[1] 744102   6639

$PBMC
[1] 744102   9939

Repeat here on the combined object

cr.combined <- FindTopFeatures(cr.combined, min.cutoff = 'q0')
cr.combined <- RunTFIDF(cr.combined, assay="ATAC")
Performing TF-IDF normalization
cr.combined <- RunSVD(cr.combined)
Running SVD
Scaling cell embeddings
cr.combined <- RunUMAP(cr.combined, reduction = "lsi", dims = 2:30)
06:18:50 UMAP embedding parameters a = 0.9922 b = 1.112
06:18:50 Read 16578 rows and found 29 numeric columns
06:18:50 Using Annoy for neighbor search, n_neighbors = 30
06:18:50 Building Annoy index with metric = cosine, n_trees = 50
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
06:18:54 Writing NN index file to temp file /tmp/RtmpwGLNhI/file3b98f15031c5bd
06:18:54 Searching Annoy index using 1 thread, search_k = 3000
06:19:04 Annoy recall = 100%
06:19:09 Commencing smooth kNN distance calibration using 1 thread
06:19:16 Initializing from normalized Laplacian + noise
06:19:18 Commencing optimization for 200 epochs, with 670814 positive edges
06:19:35 Optimization finished

Integrate all data

# Find integration anchors
integration.anchors <- FindIntegrationAnchors(object.list = cr.list, 
    anchor.features = rownames(cr.combined), 
    reduction = "rlsi", 
    dims = 2:30
)
Computing within dataset neighborhoods
Finding all pairwise anchors
Warning: No filtering performed if passing to data rather than counts
Projecting new data onto SVD
Projecting new data onto SVD
Finding neighborhoods
Finding anchors
    Found 154 anchors
# Integrate LSI embeddings
# Arbitrarily reduced k.weight to 50 to get around "number of items to replace is not a multiple of replacement length" error observed elsewhere
cellranger.integrated <- IntegrateEmbeddings(anchorset = integration.anchors, 
    reductions = cr.combined[["lsi"]],
    new.reduction.name = "integrated_lsi",
    dims.to.integrate = 1:30,
    k.weight = 50
)
Merging dataset 1 into 2
Extracting anchors for merged samples
Finding integration vectors
Finding integration vector weights
Integrating data

Create a new UMAP using the integrated embeddings

cellranger.integrated <- RunUMAP(cellranger.integrated, reduction = "integrated_lsi", dims = 2:30)
06:30:28 UMAP embedding parameters a = 0.9922 b = 1.112
06:30:28 Read 16578 rows and found 29 numeric columns
06:30:28 Using Annoy for neighbor search, n_neighbors = 30
06:30:28 Building Annoy index with metric = cosine, n_trees = 50
0%   10   20   30   40   50   60   70   80   90   100%
[----|----|----|----|----|----|----|----|----|----|
**************************************************|
06:30:31 Writing NN index file to temp file /tmp/RtmpwGLNhI/file3b98f113e5debd
06:30:31 Searching Annoy index using 1 thread, search_k = 3000
06:30:41 Annoy recall = 100%
06:30:49 Commencing smooth kNN distance calibration using 1 thread
06:30:56 Initializing from normalized Laplacian + noise
06:30:58 Commencing optimization for 200 epochs, with 670716 positive edges
06:31:14 Optimization finished
DimPlot(cellranger.integrated, group.by = "Sample")

Identify clusters

cellranger.integrated <- FindNeighbors(object = cellranger.integrated, reduction = 'integrated_lsi', dims = 2:30)
Computing nearest neighbor graph
Computing SNN
cellranger.integrated <- FindClusters(object = cellranger.integrated, verbose = FALSE, algorithm = 3, resolution = 0.5)
DimPlot(object = cellranger.integrated, label = TRUE) + NoLegend()

Compute gene activity matrix and insert this as a pseudo RNA assay

Annotation(cellranger.integrated) <- annotations
gene.activities <- GeneActivity(cellranger.integrated)
Extracting gene coordinates
Extracting reads overlapping genomic regions
Extracting reads overlapping genomic regions
# Add the gene activity matrix to the Seurat object as a new assay and normalize it
cellranger.integrated[['RNA']] <- CreateAssayObject(counts = gene.activities)
cellranger.integrated <- NormalizeData(
  object = cellranger.integrated,
  assay = 'RNA',
  normalization.method = 'LogNormalize',
  scale.factor = median(cellranger.integrated$nCount_RNA)
)

DefaultAssay(cellranger.integrated) <- 'RNA'

Plot the same marker genes from Seurat vignette

FeaturePlot(
  object = cellranger.integrated,
  features = c('MS4A1', 'CD3D', 'LEF1', 'NKG7', 'TREM1', 'LYZ'),
  pt.size = 0.1,
  max.cutoff = 'q95',
  ncol = 3
)

Save workspace

save.image("Cellranger_HGMM_PBMC_seurat_090222_QC_integrated.RData")

sessionInfo()
R version 4.1.0 (2021-05-18)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Red Hat Enterprise Linux 8.5 (Ootpa)

Matrix products: default
BLAS/LAPACK: /nas/longleaf/rhel8/apps/r/4.1.0/lib/libopenblas_haswellp-r0.3.5.so

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [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] stats4    stats     graphics  grDevices utils     datasets  methods  
[8] base     

other attached packages:
 [1] EnsDb.Hsapiens.v86_2.99.0 ensembldb_2.18.3         
 [3] AnnotationFilter_1.18.0   GenomicFeatures_1.46.5   
 [5] AnnotationDbi_1.56.2      Biobase_2.54.0           
 [7] Signac_1.7.0.9003         SeuratObject_4.0.4       
 [9] Seurat_4.1.0              GenomicRanges_1.46.1     
[11] GenomeInfoDb_1.30.1       IRanges_2.28.0           
[13] S4Vectors_0.32.4          BiocGenerics_0.40.0      
[15] forcats_0.5.1             stringr_1.4.0            
[17] dplyr_1.0.9               purrr_0.3.4              
[19] readr_2.1.2               tidyr_1.2.0              
[21] tibble_3.1.8              ggplot2_3.3.6            
[23] tidyverse_1.3.1           workflowr_1.7.0          

loaded via a namespace (and not attached):
  [1] utf8_1.2.2                  reticulate_1.25            
  [3] tidyselect_1.1.2            RSQLite_2.2.10             
  [5] htmlwidgets_1.5.4           grid_4.1.0                 
  [7] BiocParallel_1.28.3         Rtsne_0.15                 
  [9] munsell_0.5.0               codetools_0.2-18           
 [11] ica_1.0-2                   future_1.24.0              
 [13] miniUI_0.1.1.1              withr_2.5.0                
 [15] spatstat.random_2.1-0       colorspace_2.0-3           
 [17] filelock_1.0.2              highr_0.9                  
 [19] knitr_1.37                  rstudioapi_0.13            
 [21] ROCR_1.0-11                 tensor_1.5                 
 [23] listenv_0.8.0               labeling_0.4.2             
 [25] MatrixGenerics_1.6.0        git2r_0.30.1               
 [27] GenomeInfoDbData_1.2.7      polyclip_1.10-0            
 [29] farver_2.1.0                bit64_4.0.5                
 [31] rprojroot_2.0.2             parallelly_1.30.0          
 [33] vctrs_0.4.1                 generics_0.1.2             
 [35] xfun_0.30                   BiocFileCache_2.2.1        
 [37] R6_2.5.1                    DelayedArray_0.20.0        
 [39] bitops_1.0-7                spatstat.utils_2.3-0       
 [41] cachem_1.0.6                assertthat_0.2.1           
 [43] BiocIO_1.4.0                promises_1.2.0.1           
 [45] scales_1.2.0                gtable_0.3.0               
 [47] globals_0.14.0              processx_3.5.2             
 [49] goftest_1.2-3               rlang_1.0.4                
 [51] RcppRoll_0.3.0              splines_4.1.0              
 [53] rtracklayer_1.54.0          lazyeval_0.2.2             
 [55] spatstat.geom_2.3-2         broom_1.0.0                
 [57] yaml_2.3.5                  reshape2_1.4.4             
 [59] abind_1.4-5                 modelr_0.1.8               
 [61] backports_1.4.1             httpuv_1.6.5               
 [63] tools_4.1.0                 ellipsis_0.3.2             
 [65] spatstat.core_2.4-0         jquerylib_0.1.4            
 [67] RColorBrewer_1.1-3          ggridges_0.5.3             
 [69] Rcpp_1.0.8.3                plyr_1.8.7                 
 [71] progress_1.2.2              zlibbioc_1.40.0            
 [73] RCurl_1.98-1.6              prettyunits_1.1.1          
 [75] ps_1.6.0                    rpart_4.1.16               
 [77] deldir_1.0-6                pbapply_1.5-0              
 [79] cowplot_1.1.1               zoo_1.8-9                  
 [81] SummarizedExperiment_1.24.0 haven_2.4.3                
 [83] ggrepel_0.9.1               cluster_2.1.2              
 [85] fs_1.5.2                    magrittr_2.0.2             
 [87] RSpectra_0.16-0             data.table_1.14.2          
 [89] scattermore_0.8             lmtest_0.9-40              
 [91] reprex_2.0.1                RANN_2.6.1                 
 [93] whisker_0.4                 ProtGenerics_1.26.0        
 [95] fitdistrplus_1.1-6          matrixStats_0.62.0         
 [97] hms_1.1.1                   patchwork_1.1.1            
 [99] mime_0.12                   evaluate_0.15              
[101] xtable_1.8-4                XML_3.99-0.9               
[103] readxl_1.3.1                gridExtra_2.3              
[105] biomaRt_2.50.3              compiler_4.1.0             
[107] KernSmooth_2.23-20          crayon_1.5.1               
[109] htmltools_0.5.2             mgcv_1.8-40                
[111] later_1.3.0                 tzdb_0.2.0                 
[113] lubridate_1.8.0             DBI_1.1.2                  
[115] dbplyr_2.1.1                rappdirs_0.3.3             
[117] MASS_7.3-55                 Matrix_1.4-0               
[119] cli_3.3.0                   parallel_4.1.0             
[121] igraph_1.3.3                pkgconfig_2.0.3            
[123] GenomicAlignments_1.30.0    getPass_0.2-2              
[125] plotly_4.10.0               spatstat.sparse_2.1-0      
[127] xml2_1.3.3                  bslib_0.3.1                
[129] XVector_0.34.0              rvest_1.0.2                
[131] callr_3.7.0                 digest_0.6.29              
[133] sctransform_0.3.3           RcppAnnoy_0.0.19           
[135] spatstat.data_2.1-2         Biostrings_2.62.0          
[137] rmarkdown_2.12              cellranger_1.1.0           
[139] leiden_0.3.9                fastmatch_1.1-3            
[141] uwot_0.1.11                 restfulr_0.0.13            
[143] curl_4.3.2                  shiny_1.7.1                
[145] Rsamtools_2.10.0            rjson_0.2.21               
[147] lifecycle_1.0.1             nlme_3.1-155               
[149] jsonlite_1.8.0              viridisLite_0.4.0          
[151] fansi_1.0.3                 pillar_1.7.0               
[153] lattice_0.20-45             KEGGREST_1.34.0            
[155] fastmap_1.1.0               httr_1.4.2                 
[157] survival_3.2-13             glue_1.6.2                 
[159] png_0.1-7                   bit_4.0.4                  
[161] stringi_1.7.6               sass_0.4.0                 
[163] blob_1.2.2                  memoise_2.0.1              
[165] irlba_2.3.5                 future.apply_1.8.1