Question: DiffBind minOverlap parameter
0
gravatar for mademoisElle
27 days ago by
mademoisElle0 wrote:

Hi, I'm trying to use DiffBind for my ChIPseq analysis, but I encountered the following "problem".

I loaded two patients' groups of samples using dba function and setting minOverlap equal to 2. Same setting also for:

my_peaks <- dba.count(my_peaks, bUseSummarizeOverlaps=F,minOverlap=2).

I defined the contrast and then I ran a differential analysis: my_peaks <- dba.analyze(my_peaks,method=DBA_DESEQ2)

But when I looked into this object: peaks_DF <- dba.report( my_peaks,bCalled=TRUE,th=0.1) I found some peaks present only in 1 one sample for the groupA, and only in one sample for the groupB.

Is there any possibility to avoid this and having only a minOverlap of 2 in each patients' group? I mean at least a minimum of overlap=2 in each considered category (>=2 in the groupA and >=2 in the group B, not >=2 groupA+groupB) to create a consensus on which perform differential analysis.

Thank you in advance for your help!

ADD COMMENTlink modified 27 days ago by Rory Stark2.9k • written 27 days ago by mademoisElle0
Answer: DiffBind minOverlap parameter
0
gravatar for Rory Stark
27 days ago by
Rory Stark2.9k
CRUK, Cambridge, UK
Rory Stark2.9k wrote:

You can make the consensus peakset in two steps: first making a consensus for each sample group with minOverlap=2, then taking the union of these with minOverlap=1.

Here's an example using the sample data, where we get the peaks that overlap in at least two of the Responsive samples and at least two of the Resistant samples, the count using all of these peaks:

# Load peak data
data(tamoxifen_peaks)

# Generate consensus for each condition
tamoxifen.consensus <- dba.peakset(tamoxifen,consensus = DBA_CONDITION, minOverlap=2)

# Examine sample groups consensus peaks and overlap
dba.show(tamoxifen.consensus,tamoxifen.consensus$masks$Consensus)
dba.plotVenn(tamoxifen.consensus,tamoxifen.consensus$masks$Consensus)

# Retrieve union of consensus peaks
consensus <- dba.peakset(tamoxifen.consensus, bRetrieve=TRUE,
                         peaks=tamoxifen.consensus$masks$Consensus,
                         minOverlap=1)

# Count using pre-computed consensus peakset
tamoxifen <- dba.count(tamoxifen, peaks=consensus)
ADD COMMENTlink modified 27 days ago • written 27 days ago by Rory Stark2.9k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 16.09
Traffic: 437 users visited in the last hour