Hi @Saurabh Mishra?, for those who are coming from high performance computing, there was a webinar to learn about equivalent cloud-based methods for running bioinformatics jobs quickly on the Research Analysis Platform.
I have watched that webinar. I wondered if I could because my R script is taking more time to run in SAK ( even more time than the R studio workbench). So I thought I might try PBS as well.
Thanks @Chai Fungtammasan?. I use the webinar commands only to run my script in R. Why is the same R script slower in SAK than in Rstudio, even after using higher instances?
Yes, I am talking about the same discussion. As you suggested:
I ran the R script in SAK. It worked and was cost-efficient but not time efficient.
You suggested learning about docker images as some packages took a lot of time to load each time I ran. So I learned it and created an image using ttyd. But the same R script which I used on Rstudio and SAK, is throwing me the following error:
I have used system("dx download file ") in my R script.
I can't think of any reasons, but here are some suggestions you could try.
1) make sure the slowness came from running Rscript itself, not the other component. For example, you can use dxfuse or dx download in both Rstudio and SAK. Just make sure you use the same thing.
2) You can install R in cloud_workstation or ttyd and check if you script would also run slower there.
Comments
8 comments
Hi @Saurabh Mishra?, for those who are coming from high performance computing, there was a webinar to learn about equivalent cloud-based methods for running bioinformatics jobs quickly on the Research Analysis Platform.
Please see the following post: https://community.dnanexus.com/s/question/0D5t000003rH4jnCAC/webinar-may-12-cloud-computing-for-hpc-users
Equivalent commands are on the slide #16, but I would recommend to watch the whole recording as it presents many useful information.
I have watched that webinar. I wondered if I could because my R script is taking more time to run in SAK ( even more time than the R studio workbench). So I thought I might try PBS as well.
The slower R in SAK is a separated issue, and I think I answer on this topic before.
The PBS script is not supported on cloud since it was designed for HPC, so you would have to concert it using guild in webinar that Ondrej shared.
Thanks @Chai Fungtammasan?. I use the webinar commands only to run my script in R. Why is the same R script slower in SAK than in Rstudio, even after using higher instances?
Is it the installation issue that we discussed here? https://community.dnanexus.com/s/question/0D582000000aWa0CAE/how-to-run-the-r-script-faster-on-ukb-rap
Yes, I am talking about the same discussion. As you suggested:
I have used system("dx download file ") in my R script.
I can't think of any reasons, but here are some suggestions you could try.
1) make sure the slowness came from running Rscript itself, not the other component. For example, you can use dxfuse or dx download in both Rstudio and SAK. Just make sure you use the same thing.
2) You can install R in cloud_workstation or ttyd and check if you script would also run slower there.
Okay, Thanks @Chai Fungtammasan? and @Ondrej Klempir? for all the support.
Please sign in to leave a comment.