Conda Env for indexForSailfish error (libboost)

Question

Conda Env for indexForSailfish error (libboost)

Closed this issue 5 years ago · 10 comments

The "sailfish" conda environment is also used for the function indexForSailfish. However indexForSailfish requires libboost=1.60.0, but on the "sailfish" conda environment it's 1.70.0 so it's throwing an error. Reverting back to the "cgat-flow" environment (libboost=1.66.0) appears to work.

Answer 1 · 2019-11-08T07:36:25.000Z

Many thanks, I will take a look asap.

Answer 2 · 2019-11-08T07:51:24.000Z

(it works up to a point - and then fails with an m-alloc error, but that is a problem with our cluster that is happening increasingly frequently)

Answer 3 · 2019-11-08T09:09:03.000Z

Iv has earlier versions of salmon failing because of bad memory allocation on the cluster before. I think there was memory leaks on the cluster as this didn’t happen when I ran it locally. What’s your rationale sailfish and not salmon?

Answer 4 · 2019-11-08T09:41:35.000Z

No particular reason for sailfish other than it's part of this pipeline, and am trying to get it running again. Salmon works fine. I am just flagging this up, happy to close issue if no appetite in supporting the pipeline as it stands. Obviously no need to run sailfish and could be refactored towards salmon.

Answer 5 · 2019-11-08T10:22:48.000Z

My preference would be to switch to salmon and remove sailfish. I like the pipeline and would continue to support it, but I dont have many RNA-seq datasets at the moment so I haven't used it for a while. I am happy to debug on my end when I have time (Maybe next week as im at a conference and some sessions dont look as interesting).

Answer 6 · 2019-11-09T18:48:02.000Z

Could you please confirm whether it's worth fixing the issue with sailfish then?

Answer 7 · 2019-11-10T07:46:52.000Z

I personally think salmon is the way to go because sailfish isn't really supported anymore.

Answer 8 · 2019-11-13T09:21:28.000Z

Sailfish is being installed/used in its own conda environment:

https://github.com/cgat-developers/cgat-flow/blob/master/conda/environments/pipelines-sailfish.yml

Right now we use the latest available version but we could pin sailfish or its dependencies to ensure it works.

Do you want me to investigate further? Or are you finally getting rid of sailfish?

Answer 9 · 2019-11-13T09:25:13.000Z

Getting rid of sailfish. No need to support both, last sailfish release 2016. Jakub On 13 Nov 2019, at 09:21, Sebastian Luna-Valero <notifications@github.com<mailto:notifications@github.com>> wrote: Sailfish is being installed/used in its own conda environment: https://github.com/cgat-developers/cgat-flow/blob/master/conda/environments/pipelines-sailfish.yml Right now we use the latest available version but we could pin sailfish or its dependencies to ensure it works. Do you want me to investigate further? Or are you finally getting rid of sailfish? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub<#113?email_source=notifications&email_token=AC4Y5MVXNZHNSIEEL6Y44R3QTPBJTA5CNFSM4JKTU5YKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOED5OJNY#issuecomment-553313463>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AC4Y5MRMMCQ5MUSLG3I7JADQTPBJTANCNFSM4JKTU5YA>.

Answer 10 · 2019-11-13T09:28:01.000Z

Thanks for confirming, Jakub.

Please bear in mind that getting rid of existing dependencies will help us reduce the number and complexity of conda environments, which is one of the bottlenecks of the continuous integration testing. If you guys are not comfortable modifying the conda environment, please feel free to ping me and I will happily look into that for you.

Best regards,
Sebastian