Hello I have a question regarding DiffBind analyzing of ATACseq data . I have ATACseq counts of control and case samples in a reference peak set as a count matrix where row are regions and columns are sample IDs,I do not have access to bam files nor peak calls of these samples . Is it possible to use DiffBind to identify regions that are differentially accessible between case and control ? if so how can I construct a DBA object from these count matrices ?

Yes you can do this (assuming you have the region coordinates, corresponding to the rows, in chromosome-start-end form.

There are two main ways to create the DBA object. You can do it with successive calls to dba.peakset(), adding the counts for each sample. For example:

data(tamoxifen_counts)
tamoxifen <- dba.count(tamoxifen, peaks=NULL, score=DBA_SCORE_READS)
counts <- dba.peakset(tamoxifen, bRetrieve = TRUE)

newDBA <- NULL
countMatrix <- mcols(counts)
for(sample in 1:ncol(countMatrix)) {
  newDBA <- dba.peakset(newDBA, peaks=counts, 
                        sampID=colnames(countMatrix)[sample],
                        counts=countMatrix[,sample])
}  

newDBA <- dba.contrast(newDBA,group1=c(1:2,8:9), group2=c(3:7,10:11))
dba.analyze(newDBA, bBlacklist = FALSE)

You can also add metadata by setting the tissue, factor, condition, treatment, and/or replicate parameters.

If you want to use a samplesheet to load the project using dba(), you can write each column of the count matrix to a separate file with four tab- or comma-delimited columns: chromosome, start, end, counts. The first three columns must be identical for all samples. The include these files in your samplesheet in a column headed Counts.

Regarding the performance, make sure you are on the very latest version (DiffBind_3.2.4) -- there is a performance improvement that hugely speeds up adding a lot of samples. However a 100,000 interval x 400 sample matrix will take a lot of memory, so if you have limited memory it will still take a long time no matter how you do it.

Regarding your code, I see two issues:

1. The column names need to be exact (including case): SampleID, Condition
2. In the count files you are writing out, you should not include the column headers, so set col.names=FALSE.

So building on the code snippet from my previous response:

peaks <- dba.peakset(tamoxifen, bRetrieve = TRUE, DataType=DBA_DATA_FRAME)
counts1 <- cbind(peaks[,1:3],countMatrix[,1])
counts2 <- cbind(peaks[,1:3],countMatrix[,2])

write.table(counts1, "samp1", row.names=FALSE, col.names=FALSE,sep="\t", 
            quote=FALSE)
write.table(counts2, "samp2", row.names=FALSE, col.names=FALSE,sep="\t", 
            quote=FALSE)

newDBA <- dba.peakset(NULL, peaks=peaks, sampID="samp1", 
                      condition="cond1", counts="samp1")
newDBA <- dba.peakset(newDBA, peaks=peaks, sampID="samp2", 
                      condition="cond2", counts="samp2")

newDBA