sartorlab / methylSig

R package for DNA methylation analysis
17 stars 5 forks source link

methylSigReadData bug #22

Closed JLGlass closed 4 years ago

JLGlass commented 5 years ago

When running

meth <- methylSigReadData(vec.files, pData=df.pData, assembly='hg19', destranded=T, num.cores=8)

I get the following error:

Error in bsseq::read.bismark(files = fileList, sampleNames = rownames(pData), : unused arguments (sampleNames = rownames(pData), fileType = fileType, mc.cores = num.cores)

Running this same code ~1 month ago didn't cause this issue, so I suspect it's due to an update in bsseq::read.bismark?

rcavalcante commented 5 years ago

Hi,

Your suspicion is very likely correct. There was some point at which bsseq::read.bismark dropped the sampleNames and pData arguments in favor of colData. They also dropped the fileType.

I wonder, when your code was working, were you using R 3.4.z and Bioconductor 3.6? And now that your code is broken, have you upgraded your version of R to 3.5.z and are using Bioconductor 3.8. Or some similar situation?

The current release of methylSig is only meant to work in R 3.4.z with Bioconductor 3.6 for this reason, in addition to an update to bsseq that doesn't allow the GRanges object in their BSseq class to have widths > 1, which destroys the tiling functionality of methylSig.

Long story short, if it's possible for you to go back to the R 3.4.z and Bioconductor 3.6 ecosystem, I would recommend that for now (Docker or Singularity is a nice option and I can help you with a Dockerfile if you're unfamiliar with setting that up).

Otherwise, getting methylSig to play nice with the updated bsseq versions is on my to do list (sometime in the next two weeks actually), so if time isn't of the essence, I can hopefully get it fixed for you.

Thanks for using methylSig, and for reaching out with your problem.

Raymond

JLGlass commented 5 years ago

Hi Raymond,

Thank you for the quick reply. That sounds exactly right - I updated bioconductor relatively recently. If you have a dockerfile, that would be very helpful. I'd be happy to test out the code when you've updated it.

Best,

Jacob

rcavalcante commented 5 years ago

I think this should do it:

https://github.com/rcavalcante/dockers/tree/master/bioc_release_36

You may want to edit the Bioconductor packages to make it install faster. This is the image I keep around to deal with problems in the packages I support so there's a bunch of extraneous stuff to methylSig, e.g. annotatr, chipenrich, etc.

jsl26 commented 5 years ago

Hi Raymond,

Have run into the same problem as Jacob and was wondering if you had a chance to look at the issue?

thanks

joe

bhanratt commented 4 years ago

It is now August 2019 and I am getting the same issue as the above users. Should we consider this package no longer supported going forward or will it be updated?

rcavalcante commented 4 years ago

Hello,

This is open source software and I welcome pull requests. A roadblock to updating the package was that the bsseq package dependency changed a behavior that we really relied on.

With that said, there is a docker image that I've been using when I need to use methylSig (https://hub.docker.com/r/umichbfxcore/r_344).

I do want to get methylSig working with the most current version of R, and I hope to dedicate time to that before the summer is out. Other priorities got in the way of this project.

Thanks.

sixone11 commented 4 years ago

Summer is out.... i feel so sad that it still can't be used in newer R version.

rcavalcante commented 4 years ago

Hi, this is what I'm going to work on this week.

marin-e commented 4 years ago

Hello,

I have the same issue as the above users. Will methylSig be updated soon?

Thank you.

marin-e commented 4 years ago

Hello, I successfully ran methylSig using your docker image but I failed to run methylSigTile and methylSigAnnotation because the hg38 assembly version is not supported with these versions of annotatr and GenomeInfoDb. Do you have an idea of how solving this?

Thank you

rcavalcante commented 4 years ago

Hello all,

In anticipation of submitting the package to Bioconductor, I have done some refactoring of the methylSig code base.

Please see the vignette which is included as part of the README.md for details. The package now works with R 3.6 and Bioc 3.10 as well as in the devel versions of Bioc.

Please give it a try. For now, I will close this issue. Please open a new issue with problems relating to the new release (https://github.com/sartorlab/methylSig/releases/tag/v0.99.0).