Cannot install GenomicFeatures for R 3.6.0

[Qianhui]

Hello, I cannot install package GenomicFeatures for R 3.6.0 at RStudio Cloud, so packages including minfi and missMethyl cannot be installed. Is there a good way to install GenomicFeatures package? Thank you so much!

Killed
ERROR: lazy loading failed for package ‘GenomicFeatures’
* removing ‘/home/rstudio-user/R/x86_64-pc-linux-gnu-library/3.6/GenomicFeatures’

The downloaded source packages are in
	‘/tmp/RtmpiUgpsm/downloaded_packages’
Installation path not writeable, unable to update packages: boot, cluster, foreign, KernSmooth,
  lattice, MASS, Matrix, mgcv, nlme, survival
Warning message:
In install.packages(...) :
  installation of package ‘GenomicFeatures’ had non-zero exit status

[technocrat]

Hi, and welcome!

If you are on MacOS or Linux, the rstudio-user appears to lack write permissions for the default system-wide R package library. Two approaches are to grant the user write permission or to use sudo R to do the install.

[Qianhui]

Thank you so much, I will try these 2 approaches and get back to you soon.

[Qianhui]

Hello, sorry I forgot to mention that I am trying to install these packages on RStudio Cloud, so I don't have permission to use sudo R, is there any other way to install them? thank you so much!

[technocrat]

Ah, unfortunately, I've never used cloud. It should be possible to configure out of the issue with https://rstudio.cloud/learn/guide If you get stuck, I'll see if I can find someone to weigh in

[Qianhui]

I will have a look, thank you!

[andresrcs]

@Qianhui

Killed ERROR: lazy loading failed for package ‘GenomicFeatures’

This is happening because packages from Bioconductor need to be installed from source and compilation for this package requires more than 1GB of RAM wich exceeds the current limit for RStudio Cloud

Your only hope would be to request more memory but it is up to RStudio to decide if you have a good reason for it and if they give you the resources or not.

[Qianhui]

Thank you for your answer, I will try to request more RAM:)

[technocrat]

An alternative is to provision an R Server instance on AWS. It's said to be fairly straightforward, although I haven't tried it myself. I do know, however, that once you have the workflow of provision-configure-run-quit down it's quite economical to use EC2.

[Qianhui]

Thank you, your suggestions are very helpful!

[andresrcs]

@technocrat

it's quite economical to use EC2.

It is actually free for a year if you limit yourself to the resources included in the free tier but if you follow this road, be careful not to play with options you don't fully understand because you can end up with a nasty surprise on your credit card bill.