Question: Using R Studio on Cloud
0
gravatar for lokraj2003
4 months ago by
lokraj200390
lokraj200390 wrote:

I don't know if it is the right platform to ask this question. But, I could not find any better place than this to ask this question.

I am trying to do some single-cell analysis for a group project. I have about 500 MB of data (count matrixes). I am not able to run any analysis on my PC where I have 8GB RAM. Then I tried to use rstudio.connect. I could not even upload all the files there. The session gets disconnected within a few minutes. Then, I bought a Rstudio Droplet from Digital Ocean. It is worse than rstudio cloud. I could use HPC from my university then I won't be able to run any Rmd files. I tried to use docker on HPC, but I don't have permission to create a docker image. So, my question is what other options are available for working with moderate size data on R Studio. I would not mind paying a few bucks?

Thanks

single cell rstudio rna-seq R • 175 views
ADD COMMENTlink modified 4 months ago by zx87549.7k • written 4 months ago by lokraj200390

I could use HPC from my university then I won't be able to run any Rmd files.

Why don't you connect to the HPC with X11-forwarding and then run RStudio on the HPC? Most schedulers allow to start interactive jobs? So you simply book a session on a node, login to the node, run RStudio there.

ADD REPLYlink written 4 months ago by ATpoint40k

Why don't you connect to the HPC with X11-forwarding and then run RStudio on the HPC?

That could work but in my experience the X11-forwarding is painfully slow (that's why I wrote this genome browser). I use R a LOT and virtually all of it on remote via ssh. Personally, I'm just happy with copying commands from a vim terminal to the R interpreter that ships with R, save plots and output to file and scp them to my local machine.

ADD REPLYlink modified 4 months ago • written 4 months ago by dariober11k

You can render .Rmd files on your university's HPC. I create the .Rmd file locally, and send it to the cluster. I have a script there called render.R that looks like this:

rmarkdown::render("path/to/uploaded/file.Rmd")

I then submit a job calling that script (my uni uses slurm, your job-submission protocol might be different):

sbatch -t 24:00:00 -c 2 --mem=80G --wrap="Rscript /path/to/render.R"

I make sure to save my single cell objects at the end of the analysis, and you get a nice .html output with plots / whatever else you want which you can then send back to your local machine and view.

ADD REPLYlink modified 4 months ago • written 4 months ago by jrleary130
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 1028 users visited in the last hour