How to get peak names/peak coordinates from the PeakMatrix?

[RegnerM2015]

Hi @jgranja24 and @rcorces,

I am trying to find the peak names or peak coordinates for the PeakMatrix. When I use getPeakSet(), the peak indices are NOT in the same order despite having the same number of peak indices:

peaks.mat <- getMatrixFromProject(atac,useMatrix = "PeakMatrix")
peaks <- getPeakSet(atac)

The dimensions are consistent but the order is not:

> length(peaks)
[1] 519525
> dim(peaks.mat)
[1] 519525  72507
> length(which(peaks.mat@elementMetadata$idx == peaks@elementMetadata$idx))
[1] 86173

How can I find the correct order of peak names? Thanks!

[rcorces]

if you look at rowRanges(peaks.mat), they are still the same peaks but somehow they get shuffled. Just re-order them based off chr, start, and stop.
something like this:

peaks_label <- paste0(seqnames(peaks),"_", start(ranges(peaks)),"_", end(ranges(peaks)))
peaks_reorder <- peaks[order(peaks_label)]
peaks.mat_label <- paste0(seqnames(rowRanges(peaks.mat)),"_", start(ranges(rowRanges(peaks.mat))),"_", end(ranges(rowRanges(peaks.mat))))
peaks.mat_reorder <- peaks.mat[order(peaks.mat_label)]
all.equal(start(ranges(rowRanges(peaks.mat_reorder))), start(ranges(peaks_reorder)))

[RegnerM2015]

@rcorces and @jgranja24

Would this shuffling of peak labels in the getPeakSet() function also affect the addPeak2GeneLinks() procedure? In other words, I want to make sure I'm not correlating with the wrong peak rows from the peak matrix.

[rcorces]

The peak shuffling is caused by making the matrix. I dont think it affects addPeak2GeneLinks().

Also, I'm not sure how your peaks.mat object lacks rowRanges.

> peaks.mat <- getMatrixFromProject(ArchRProj = projHeme5, useMatrix = "PeakMatrix")
ArchR logging to : ArchRLogs/ArchR-getMatrixFromProject-365f1318bde5-Date-2021-06-22_Time-08-14-40.log
If there is an issue, please report to github with logFile!
2021-06-22 08:14:44 : Organizing colData, 0.064 mins elapsed.
2021-06-22 08:14:44 : Organizing rowData, 0.064 mins elapsed.
2021-06-22 08:14:44 : Organizing rowRanges, 0.064 mins elapsed.
2021-06-22 08:14:44 : Organizing Assays (1 of 1), 0.064 mins elapsed.
2021-06-22 08:14:44 : Constructing SummarizedExperiment, 0.067 mins elapsed.
2021-06-22 08:14:45 : Finished Matrix Creation, 0.084 mins elapsed.
> rowRanges(peaks.mat)
GRanges object with 143400 ranges and 1 metadata column:
           seqnames              ranges strand |     idx
              <Rle>           <IRanges>  <Rle> | <array>
       [1]     chr1       752502-753002      * |       1
       [2]     chr1       762685-763185      * |       2
       [3]     chr1       779917-780417      * |       3
       [4]     chr1       793283-793783      * |       4
       [5]     chr1       801006-801506      * |       5
       ...      ...                 ...    ... .     ...
  [143396]     chrX 154840785-154841285      * |    3413
  [143397]     chrX 154841356-154841856      * |    3414
  [143398]     chrX 154842430-154842930      * |    3415
  [143399]     chrX 154862053-154862553      * |    3416
  [143400]     chrX 154997960-154998460      * |    3417
  -------
  seqinfo: 23 sequences from an unspecified genome; no seqlengths

[RegnerM2015]

@rcorces

Can you confirm that rowRanges is from the SummarizedExperiment library? Or is rowRanges() from another library?

> SummarizedExperiment::rowRanges(peaks.mat)
NULL

[rcorces]

correct.

> SummarizedExperiment::rowRanges(peaks.mat)
GRanges object with 143400 ranges and 1 metadata column:
           seqnames              ranges strand |     idx
              <Rle>           <IRanges>  <Rle> | <array>
       [1]     chr1       752502-753002      * |       1
       [2]     chr1       762685-763185      * |       2
       [3]     chr1       779917-780417      * |       3
       [4]     chr1       793283-793783      * |       4
       [5]     chr1       801006-801506      * |       5
       ...      ...                 ...    ... .     ...
  [143396]     chrX 154840785-154841285      * |    3413
  [143397]     chrX 154841356-154841856      * |    3414
  [143398]     chrX 154842430-154842930      * |    3415
  [143399]     chrX 154862053-154862553      * |    3416
  [143400]     chrX 154997960-154998460      * |    3417
  -------
  seqinfo: 23 sequences from an unspecified genome; no seqlengths

[RegnerM2015]

@rcorces

Ah. I see now that the 0.9.5 version of ArchR does not return rowRanges

from the source code of getMatrixFromProject() v0.9.5:

se <- SummarizedExperiment(assays = asy, colData = cD, rowData = rD1)
    rm(seL)
    gc()
    .logDiffTime("Finished Matrix Creation", t1 = tstart, verbose = verbose, 
        logFile = logFile)
    se

from the source code of getMatrixFromProject(0 v1.0.1:

if(!is.null(rR1)){
    se <- SummarizedExperiment(assays = asy, colData = cD, rowRanges = rR1)
  }else{
    se <- SummarizedExperiment(assays = asy, colData = cD, rowData = rD1)
  }

I will edit the source code on my end so that I can keep using v0.9.5 for reproducibility purposes.