some help with DiffBind
1
0
Entering edit mode
jvdelarosa • 0
@jvdelarosa-9907
Last seen 8.0 years ago

Hi  everyone,

I'm trying to make a di fferential binding analysis of ChIP-Seq peak data with DiffBind. I´m trying to use the code explained in the post: A: question about DiffBind, because I don't have any replicate for my samples, and I know this is not the best way to do the analysis. Anyway, when I try to run the script, I get different warning and error messages.  

I have 3 samples with chip-seq experiment,they are TFa, TFb and KO, and 3 peak files for every sample from MACS2 in bed format, and an Input sample.  

I'm trying to remove all the peaksets that overlap with the KO sample, and obtain the overlapping and non overlapping peaksets between TFa and TFb.

Here I paste the output from the console:

 

> library(DiffBind)
>
> samples <- read.csv2("Samples.csv")

> read.csv2("Samples.csv")
  SampleID    Tissue Factor  Condition Treatment Replicate bamReads ControlID bamControl         Peaks
1    CELL1 CELL1_TFa    TFa Responsive    LIGAND         1  TFa.bam     INPUT  INPUT.bam MACS2_TFa.bed
2    CELL2 CELL2_TFb    TFb Responsive    LIGAND         1  TFb.bam     INPUT  INPUT.bam MACS2_TFb.bed
3    CELL3  CELL3_KO     KO  Resistant    LIGAND         1   KO.bam     INPUT  INPUT.bam  MACS2_KO.bed
  PeakCaller
1        bed
2        bed
3        bed
>
> diff <- dba(sampleSheet=samples)
CELL1 CELL1_TFa TFa Responsive LIGAND 1 bed
CELL2 CELL2_TFb TFb Responsive LIGAND 1 bed
CELL3 CELL3_KO KO Resistant LIGAND 1 bed
>
> diff_count <- dba.count(diff,minOverlap=1)
>
> diff_analysis <- dba.contrast(diff_count, group1=1, name1="TFa", group2=2, name2="TFb")
> diff_analysis <- dba.contrast(diff_analysis, group1=1, name1="TFa", group2=3, name2="KO")
> diff_analysis <- dba.contrast(diff_analysis, group1=2, name1="TFb", group2=3, name2="KO")
>
> diff_analysis <- dba.analyze(diff_analysis,method=DBA_DESEQ2)
Warning messages:
1: Some groups have no replicates. Results may be unreliable.
2: In dba.analyze(diff_analysis, method = DBA_DESEQ2) :
  No correlation heatmap plotted -- contrast 1 does not have enough differentially bound sites.
>
> rep1 <- dba.report(diff_analysis, contrast=1, th=1, bCount=TRUE)
Error in pv.DBAreport(pv = DBA, contrast = contrast, method = method,  :
  edgeR analysis has not been run for this contrast
> rep2 <- dba.report(diff_analysis, contrast=2, th=1, bCount=TRUE)
Error in pv.DBAreport(pv = DBA, contrast = contrast, method = method,  :
  edgeR analysis has not been run for this contrast
> rep3 <- dba.report(diff_analysis, contrast=3, th=1, bCount=TRUE)
Error in pv.DBAreport(pv = DBA, contrast = contrast, method = method,  :
  edgeR analysis has not been run for this contrast
> read.csv2("Samples.csv")

I hope somebody could help me

Thanks in advance.

Vladimir

diffbind • 1.9k views
ADD COMMENT
1
Entering edit mode
Gord Brown ▴ 650
@gord-brown-5664
Last seen 3.2 years ago
United Kingdom

Hi,

You ran your analysis with DESeq2, but the default for dba.report is to return an EdgeR analysis.  Add the argument "method=DBA_DESEQ2" to the dba.report call to get a DESeq2 results.

Cheers,

 - Gord

ADD COMMENT
0
Entering edit mode

Without replicates, it is unlikely that you will get FDR scores indicating high confidence that any sites are differentially bound. As Gord suggests, specifying the method in dba.report() and setting th=1 will let you see the fold changes, which you can use to rank the sites.  You could also set fold=2, for example, to identify the sites with a large fold change. If you also set bCalled=TRUE, you can see if the site was called as a peak by MACS as well. 

-R

ADD REPLY

Login before adding your answer.

Traffic: 534 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