Creating fragment object for common peak set (multiome 10x)

[NadiaEdel]

Dear Signac team,

Thank you for this great resource!
I apologise if this has been asked before (I found a similar thread and I could not find it again), or if this is a rather silly question, but I am stuck and I would like to be able to move forward with the analysis.

I am working with 10x multiome samples (RNA+ATAC on same nuclei). I am trying to create a common peak set following the https://stuartlab.org/signac/articles/merging tutorial. I was able to create the common peak set. However, I am stuck on the "Create Fragment objects" as I cannot find the "singlecell.csv"-type of file in the "outs" generated by CellRangerARC. I have the .bed files for the peaks, the tsv for the fragments, but I do not see any file with .csv, the closest thing I see is the "per_barcode_metrics" xls file. Could you help me?
Would it be wrong to continue without adding the metadate information?

Would this be wrong for creating a common peak set? (working in Rstudio on a Windows OS):

Create Seurat object

counts = Read10X_h5(paste0(sample, "/outs/filtered_feature_bc_matrix.h5"))
atac_fragments = paste0(sample, "/outs/atac_fragments.tsv.gz")

chrom_assay = CreateChromatinAssay(
counts=counts$Peaks,
sep=c(":","-"),
annotation = annotation_cjac,
fragments = atac_fragments,
validate.fragments=T)

seur = CreateSeuratObject(
counts = chrom_assay,
assay = "ATAC")

Merge samples

merged_seur <- merge(seu03, y = c(seu04, seu05, seu07, seu08, seu09), add.cell.ids = c("s03", "s04", "s05", "s07", "s08", "s09"))
merged_seur

#Create fragment files for each sample
frag3 <- CreateFragmentObject("/outs/atac_fragments.tsv.gz")
frag4 <- CreateFragmentObject("/outs/atac_fragments.tsv.gz")
(...)

Make new counts objects based on the combined file features

DefaultAssay(merged_seur) <- "ATAC"
counts3 <- FeatureMatrix(fragments = frag3, features = granges(merged_seur))
(...)

atac_assay3 = CreateChromatinAssay(
counts=counts3,
sep=c(":","-"),
annotation = annotation,
fragments = frag3,
validate.fragments=T)
(...)

Finally make the seurat objects

ATAC_3 <- CreateSeuratObject(counts = atac_assay3, assay = "CommonPeaks")
ATAC_4 <-
(...)

Merge samples after re-calculating ATAC peaks

atac_merged <- merge(ATAC_3, y = c(ATAC_4, ATAC_5, ATAC_7, ATAC_8, ATAC_9), add.cell.ids = c("s03", "s04", "s05", "s07", "s08", "s09"))
atac_merged

I am also confused as I have seen different options where the basic QC is done first, and then the common peak set is created, or the other way around...first create the common peak set, re-calculate based on the common peak set, and then do the basic QC. For example, here https://stuartlab.org/signac/1.11.0/articles/pbmc_multiomic, when performing the "peak calling" with MACS2 this happens after the QC. Is there a recommended approach?

Thank you in advance for your help and I apologise for any inconvenience.

Kind regards,

Nadia

[timoast]

Hi Nadia, the singlecell.csv file is not required, it just contains some metadata generated by cellranger that may be useful. The per_barcode_metrics.csv file from cellranger-arc is similar and you might want to look at data in that file to help QC your cells.

The singlecell.csv file is used in the merging vignette to identify what cells to include in the analysis, but you can substitute any method of choosing the cells to include in that step.

[NadiaEdel]

Dear Tim,

Thank you so much for taking the time to answer my post. I was able to move forward, I ended up using the "colnames" for the "cells" parameter when doing the count quantification in each peak of the common peak set following: https://stuartlab.org/signac/1.11.0/articles/pbmc_multiomic, and I was able to follow through.

I am still not fully convienced because I used the peaks from cellrangerARC, not from MACS2 (I still cannot make it work, even from the WSL Ubuntu setup). I read in another post that -I think it was you- commented that the algorithm from CellRangerARC was updated so maybe now MACS2 and CellRangerARC would give out comparable peak info, is this so? (this post was from late 2023, here: #1490).

For now the order I followed was: create common peak set, re-quantif peaks against that for each sample, do basic QC on the ATAC (also performed QC in the RNA assay separately), annotate cell types on ATAC assay from RNA cell type info.

Apologies again, I know these questions are probaly rather basic.

Thank you for your help.

Kind regards,

Nadia

[timoast]

Sounds fine to me