DiffBind getting Fold values of near 0
0
0
Entering edit mode
Miquel • 0
@4dca5ab7
Last seen 12 months ago
Austria

Hello,

I'm trying to run DiffBind on my CUT&RUN data to get the differentially bound peaks between two conditions (DMSO and inhibitor). In sample_sheet.csv I only have the four samples I want to compare, two replicates each, and I'm running the following analysis:

samples_file=read.csv("sample_sheet.csv", header=T)
Diff_dba=dba(sampleSheet=samples_file)
Diff_dba_counts=dba.count(Diff_dba, minOverlap = 2)
Diff_dba_contrast <- dba.contrast(Diff_dba_counts, categories = DBA_TREATMENT, minMembers = 2)
Diff_dba_analyze <- dba.analyze(Diff_dba_contrast)
Diff_dba_report <- dba.report(Diff_dba_analyze, file="Diffbind_results", bNormalized = TRUE, th = 1)

However, when I run it, the output table looks like this:enter image description here

The Conc values look reasonable, but then the Fold values are completelly off, near 0, which would mean a near 1 Fold change, which is clearly not what the Conc values suggest. I really don't know what the issue is here, and I would appreciate any suggestions as to what I'm doing wrong/missing. Thank you in advance!

DiffBind DESeq2 • 604 views
ADD COMMENT
1
Entering edit mode

Almost certainly data are noisy and DESeq2 (which is used internally) shrinks the fold changes towards zero, because the naive fold changes (Conc_AMG vs Conc_DMSO) have no evidence to be reliable given the sample size and noise in the data. Do PCA and MAplots to explore how the data cluster and look.

ADD REPLY

Login before adding your answer.

Traffic: 457 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6