Diffbind:dba.plotProfile error: each range must have an end that is greater or equal to its start minus one

Question

[K_QIN]

Hi,

I am using the new plotProfile function of Diffbind to create some graphs, and get the following error with one subset of my data:

dbObj$config$RunParallel <- TRUE
profiles <- dba.plotProfile(dbObj)

Generating report-based DBA object...
Generating profiles...
Error in value[[3L]](cond) : 
 profileplyr error: Error: BiocParallel errors
 element index: 2
 first error: each range must have an end that is greater or equal to its start minus one

All the rest of the ChIPQC and Diffbinds works smoothly. Also the other four samples of the same dataset worked as well (I processed them separately, as they were different treatments). What could be the problem?

Thanks, KQ2012

[pbpanigrahi]

From this Error: BiocParallel errors try dbObj$config$RunParallel <- F and run

or

check whether end >= start-1 is not satisfying?

Need more info on the input data.

[K_QIN]

I have tried the dbObj$config$RunParallel <- F also changed $core in the config, It doesn't look like it's the problem

end>= start-1 not sure how to do that?

Thanks!

Answer 1

[Rory Stark]

I know we try to avoid cross-posting, but it may be a good idea to post this issue on the Bioconductor support forum, and to tag both DiffBind and profileplyr. There's a chance that one of the profileplyr authors has seen this error and knows how to deal with it.

The advice to run this in non-parallel mode is good as sometime the error is more clear.

In the first instance, if you could send me a copy of your DiffBind object dbObj, I can have a look at what sort of profiles it is trying to generate. I may be able to spot something. If not, I'd need access to all of the data (bam files) to see where it is going wrong.

[K_QIN]

Thanks a lot! Here is the link to download the data. What I tried to do is compare TF1 or TF2 in two different tissue as indicated in the sample sheet. TF1 worked well, I was able to get the plot. One problem with TF2 is there were very few dots on one side of the volcano plot.

https://app.box.com/s/nqfeujq8rq7n0u1e2p664sjw35nlwfqz

[Rory Stark]

I had a look at the object in the link. I see that it has not gone through read counting or analysis, just the raw peaks. I'm assuming you ran into the problem after analysis, as I see it using a report-based DBA object. Could you email me a link to object after the analysis?

(Also I notice that nearly 95% of the 420k called peaks are only identified in a single sample, which may be of concern).

[K_QIN]

Hi Rory, To make it less confused, I only kept TF2 in the data. DBA object: https://app.box.com/s/tyv3a8xtumvlt972v3b94c7971jd13k6 sample sheet: https://app.box.com/s/p2tjyfb15ek0jq4wkxxfx9snqvil78mu Contrast: tissue PB vs FL

Thanks!

[Ram]

Please use GitHub gists to share datasets in plain text, and imgbb/biostars to upload images. The spambot marked your post as spam because of the box links you used and I have reverted that now, but please use better means than sharing binary files over Box links.