diffbind dba.report error

Question

[jellybean5]

Hi, I have a CUT&RUN experiment with 3 groups, 2 replicates. I ran the following:

affinity binding matrix

DBA1_3 <- dba.count(DBA1_3, bUseSummarizeOverlaps = TRUE)

normalize

DBA1_3 <- dba.normalize(DBA1_3)

Establishing contrast groups

DBA1_3 <- dba.contrast(DBA1_3, group1 = DBA1_3$masks$M, group2 = DBA1_3$masks$C, name1 = "M", name2 = "C")

DBA1_3 <- dba.contrast(DBA1_3, group1 = DBA1_3$masks$P, group2 = DBA1_3$masks$M, name1 = "P", name2 = "M")

DBA1_3 <- dba.contrast(DBA1_3, group1 = DBA1_3$masks$P, group2 = DBA1_3$masks$C, name1 = "P", name2 = "C")

DBA1_3 <- dba.analyze(DBA1_3, method=DBA_EDGER, bGreylist = FALSE)

This worked nicely when I used dba.show. However, when I tried to write dba.report for the different contrasts using:

Dreport1<- dba.report(DBA1_3, method=DBA_EDGER, contrast=1)

Dreport2 <- dba.report(DBA1_3, method=DBA_EDGER, contrast=2)

Dreport3 <- dba.report(DBA1_3, method=DBA_EDGER, contrast=3)

The first report runs okay. The other two give me the following error:

Error in .new_IRanges_from_start_width(start, width) : 'start' or 'width' cannot contain NAs

sessionInfo() gives me DiffBind version 3.4.6. Why would the first report run ok but not the other two??

Answer 1

[Rory Stark]

If I can get access to your DBA object DBA1_3, I can have a look.

Is there a particular reason you are using the old-style group1/group2 contrasts instead of the preferable method of using a design formula and specifying the conditions to test using the contrast parameter?

[jellybean5]

Thank you for your quick response. I started using the old group1/group2 I think because I started with unequal groups (n=3 for 2 groups, n=2 for one) and it kept only returning the contrast comparing the 2 groups with n=3. Since I am new here, what is the best way to give you access to the DBA object?

[Rory Stark]

You can email me at the maintainer's email address listed on the DiffBind page on the Bioconductor site. If you can save your DBA object DBA1_3 using dba.save(), store it someplace I can access it (like Dropbox), and send me the link, I can have a look.

You should be able to set the contrasts you want using a design formula. Here's an example using the sample data:

data(tamoxifen_counts)

tamoxifen <- dba.contrast(tamoxifen, 
                          contrast=c("Tissue","BT474","MCF7"))
tamoxifen <- dba.contrast(tamoxifen, 
                          contrast=c("Tissue","ZR75", "MCF7"))
tamoxifen <- dba.contrast(tamoxifen, 
                          contrast=c("Tissue","T47D", "MCF7"))

tamoxifen <- dba.analyze(tamoxifen)