rgrnastar is still cloning

Question: rgrnastar is still cloning

12 weeks ago by

I used the normal TTW method for installing new tools to install rgrnastar from iuc (d5659efd66aa) into Galaxy 18.05. I started this process yesterday afternoon, around 2 or 3.

18 hours later it still hasn't finished - I presume something has gone wrong.

Which logs would indicate where the error is, and/or how do I get it installed? Should I kill, uninstall, attempt a reinstall, or is there a better option?

rna-seq software error • 159 views

ADD COMMENT • link •

modified 12 weeks ago • written 12 weeks ago by lachlan.simpson • 20

12 weeks ago by

lachlan.simpson • 20

lachlan.simpson • 20 wrote:

Failed to install through the web again. Jumping onto the command line I was able to install successfully via bioblend and ephemeris

ADD COMMENT • link written 12 weeks ago by lachlan.simpson • 20

Ok, glad that worked. Not sure why the GUI install didn't work. Please consider giving the IUC some feedback about the install issues on their Gitter channel: https://gitter.im/galaxy-iuc/iuc. This is a somewhat new wrapper release. They would also know about existing (if any) known issues/limitations.

For the memory requirements, yes, the entire database must fit into memory during mapping.

If pre-indexed, that is the size of the genome index.
If not pre-indexed (custom genome fasta from the history), that is greatly increased - and will take longer to run. The first step in the job process will be to create the index (then discard it, these are not saved back for future use, although making that an option has been discussed). The second step in the job process is to actually run the mapping (against that new index).

If running a local, definitely create a permanent index to both to save computing time and to reduce the memory footprint, especially for large genomes you'll be using often.

So, you'll need to have enough memory to do the indexing in either case, but that would only need to be allocated once if creating a persistent index. At some later point, the public index data cache (CVMFS) will include these indexes, too, for common model organisms, but that expansion project is still under development.

ADD REPLY • link modified 12 weeks ago • written 12 weeks ago by Jennifer Hillman Jackson ♦ 25k

12 weeks ago by

Jennifer Hillman Jackson ♦ 25k

United States

Jennifer Hillman Jackson ♦ 25k wrote:

Hello,

As a first pass at troubleshooting, I would suggest uninstalling the tool then reinstalling.

This is the latest revision of the tool and it should install Ok. If you have more problems with it could you share some more information? In particular, have you installed any other tools successfully from the Tool Shed?

Let's troubleshoot more from there depending on the outcome. We may need more details about your local's configuration to get some developer input.

Thanks! Jen, Galaxy team

ADD COMMENT • link written 12 weeks ago by Jennifer Hillman Jackson ♦ 25k

12 weeks ago by

lachlan.simpson • 20

lachlan.simpson • 20 wrote:

Actually, as I read the system requirements, I think I've made a terrible mistake. Does it require - for example, with hg19 - roughly 30GB to run or does it mean 57GB?

"Memory: To run efficiently, RNA-STAR requires enough free memory to hold the SA-indexed reference genome in RAM. For Human Genome hg19 this index is about 27GB and running RNA-STAR requires approximately ~30GB of RAM. For custom genomes, the rule of thumb is to multiply the size of the reference FASTA file by 9 to estimate the required amount of RAM."

I'm running Galaxy on a server with 32GB ram and thought we could scrape in with some occasional slow downs, but a second (well, fifth or so) reading would suggest I've missed the RAM requirements by 100%

ADD COMMENT • link written 12 weeks ago by lachlan.simpson • 20

12 weeks ago by

lachlan.simpson • 20

lachlan.simpson • 20 wrote:

Thanks. I'll do that now.

ADD COMMENT • link modified 12 weeks ago • written 12 weeks ago by lachlan.simpson • 20

Please log in to add an answer.

Similar posts • Search »