Converting archr objects to seurat or singlecellexperiment objects

[sel041]

Hi archr team,

I was wondering if I can convert archr objects to seurat or singlecellexperiment objects.

Thanks!

[rcorces]

This is not currently possible. As you can imagine, the architecture of ArchR and Seurat are not super compatible. You can export certain matrices from ArchR using getMatrixFromProject() and this should allow you do get the data you want out.

[dagarfield]

See here: https://satijalab.org/seurat/archive/v3.1/conversion_vignette.html
Or just keep it as a SCE and do your analysis there: https://bioconductor.org/books/release/OSCA/

[sylestiel]

Thanks!!

[Ping-lin14]

@sylestiel

Hi! I wonder if you have successfully converted the Archr object to the Seurat object?
I tried it and I didn't succeed.
Here is my code:

proj_gs <- getMatrixFromProject(proj, useMatrix = "GeneScoreMatrix")
proj_sce <- as(proj_gs, "SingleCellExperiment")
library(scater)
proj_sce <- runPCA(proj_sce)

Error in assay(x, exprs_values) : 
  'assay(<SingleCellExperiment>, i="character", ...)' invalid subscript 'i'
'logcounts' not in names(assays(<SingleCellExperiment>))

[dagarfield]

@Ping-lin14
I'm not seeing the Seurat conversion step in your code. Possible you missed a step? In any case, you'll need to do some form of normalization before PCA, though this is outside the scope of the ArchR git.

[Ping-lin14]

@dagarfield
I think I didn't express myself clearly.
I followed https://satijalab.org/seurat/articles/conversion_vignette.html.
According to the demo, runPCA needs to be executed before executing as.Seurat.
However, I failed, I am not sure if there is one step missing.

[dagarfield]

1. runPCA doesn't (I don't think) need to be run before converstion
2. In any case, for both conversion and runPCA, you need logcounts.

Try this:

counts <- assay(proj_sce, "counts")
libsizes <- colSums(counts)
size.factors <- libsizes/mean(libsizes)
logcounts(proj_sce) <- log2(t(t(counts)/size.factors) + 1)

[pengxin2019]

Hi there,
I just tried the solution you mentioned:

proj_gs <- getMatrixFromProject(proj.doublet.filtered.harmony, useMatrix = "GeneScoreMatrix")
proj_gs

# class: SummarizedExperiment 
# dim: 30311 19265 
# metadata(0):
#   assays(1): GeneScoreMatrix
# rownames: NULL
# rowData names(6): seqnames start ... name idx
# colnames(19265): pbmc.6800.2x#GGAGAACAGGTTCGAG-1
# pbmc.6800.2x#GAACGTTCAATGTGCT-1 ... pbmc.6800.1x#GATGATCCACTCGTGG-1
# pbmc.6800.1x#TACGGATCAGGACCTC-1
# colData names(14): DoubletEnrichment DoubletScore ... TSSEnrichment
# projColData[, colnames(projColData) %ni% colnames(colData)]


BiocManager::install("SingleCellExperiment")
library(SingleCellExperiment)
proj_sce <- as(proj_gs, "SingleCellExperiment")
# class: SingleCellExperiment 
# dim: 30311 19265 
# metadata(0):
#   assays(1): GeneScoreMatrix
# rownames: NULL
# rowData names(6): seqnames start ... name idx
# colnames(19265): pbmc.6800.2x#GGAGAACAGGTTCGAG-1
# pbmc.6800.2x#GAACGTTCAATGTGCT-1 ... pbmc.6800.1x#GATGATCCACTCGTGG-1
# pbmc.6800.1x#TACGGATCAGGACCTC-1
# colData names(14): DoubletEnrichment DoubletScore ... TSSEnrichment
# projColData[, colnames(projColData) %ni% colnames(colData)]
# reducedDimNames(0):
#   altExpNames(0):


> proj_sce@assays@data
List of length 1
names(1): GeneScoreMatrix

As you can see from above, there is not countsassay. So, I am using GeneScoreMatrixas counts below:

counts <- assay(proj_sce, "GeneScoreMatrix")
# 30311 19265
counts(proj_sce) <- counts

libsizes <- colSums(counts)
size.factors <- libsizes/mean(libsizes)
logcounts(proj_sce) <- log2(t(t(counts)/size.factors) + 1)
# > proj_sce

However, its complaining that there is no rownames in the matrix when I ran as.Seurat:

library(Seurat)
#proj_sce <- runPCA(proj_sce)
proj_sce <- as.Seurat(proj_sce, counts = "counts", data = "logcounts")#

No feature names (rownames) names present in the input matrix

I have checked the arguments of getMatrixFromProject and it seems like there is no argument there enables the rownames to the
GeneScoreMatrix assay. Do you have any suggestions?

Thanks
Best

[dagarfield]

Add row names. ;)
rownames(proj_sce)<-"whatever_you_want_your_rownames_to_be"