Hi,
On my tool box I am not able to find either velveth or velvetg. Regarding velvet the only tool I can find is VelvetOptimiser Automatically optimize Velvet assemblies (Galaxy Version 2.2.6).
Can you please help with this.
Thank you
YKV
1 Like
What Galaxy are you working on (url)?
1 Like
Hi @marten,
I have found the tools(velveteg and velveth) on https://usegalaxy.eu/ . On tool column beside tool there is a caption “training only” what does that mean? Because when I used tool, run continued for two days and still going on. Do I have to consider that caption for this long run, like as it is for training purpose only?
In case if yes I would like to know how to finish De Bruijn Graph Assembly without those tools.
Thank you
Hi @YKV
These are computationally expensive tools and full-size assemblies will exceed the resources available at public Galaxy servers.
If the job is still running, allow it to complete. But if it fails for a resource-related reason, then you know the job is too large. Now, the particular server you are working at (usegalaxy.eu) will automatically resubmit failed jobs. This is great in most cases but can lead to a bit of a loop: job fails for resource > resubmits > repeat. This could happen a few times before the job finally terminates.
I suspect that is where you are now – in a loop where the job is too large but hasn’t completely finished all attempts to rerun. Reruns help when jobs fail for technical reasons (hit a problematic cluster node by chance). But if the work is actually too large, or the real error is due to a content/format issue, a rerun won’t solve the problem. The data needs to be smaller (downsampled/subsampled), or the input issue fixed, or you need to move to a Galaxy server that can handle the larger data.
In short, very large jobs require that you set up your own Galaxy and allocate sufficient resources.
- https://galaxyproject.org/choices/
- https://galaxyproject.org/use/
- My job ended with an error. What can I do? – See the options for memory and/or walltime (runtime) errors. Be sure to rule out a data input problem first. Downsampling data is also covered in the FAQ but is not always a practical (or desired) solution.