stacks upgrade - Githubissues

galaxyproject / tools-iuc

Tool Shed repositories maintained by the Intergalactic Utilities Commission

https://galaxyproject.org/iuc

MIT License

161 stars 421 forks source link

stacks upgrade #2073

Closed bernt-matthias closed 5 years ago

bernt-matthias commented 6 years ago

I guess stacks will be the next on my TODO list.

seems that there is OpenMP support which is not enabled in the Galaxy tools yet
there are newer versions 1.48 and 2.2

Would appreciate some comments before I start. @lecorguille @bgruening :)

should I go directly for stacks 2.2 .. or is this something completely new .. will also depend a bit on much time I have
bioconda contains only 2.1 so far
any experience with OpenMP?

lecorguille commented 6 years ago

ping @abretaud

@bernt-matthias There are new versions of stacks every 2 months and sometime 2 within a month so by the end of your review we should have the 2.3 or 2.5 It seems that there isn't drastic change in the UI between 1.x and 2.x but I haven't really used stacks since the 1.46 so ... No experience with OpenMP. But is it really something you need to care about?

bernt-matthias commented 6 years ago

There are new versions of stacks every 2 months and sometime 2 within a month so by the end of your review we should have the 2.3 or 2.5 It seems that there isn't drastic change in the UI between 1.x and 2.x but I haven't really used stacks since the 1.46 so ...

Good to know. Then I could go for 2.x. If there are only minor changes in upcoming versions then it should be easy to incorporate during the PR (or OK to ignore them).

No experience with OpenMP. But is it really something you need to care about?

Users were complaining about run time. Parallelization was one thing that I was thinking about.

bernt-matthias commented 6 years ago

Actually I was wrong about parallelization, its already supported by the wrappers.

lecorguille commented 6 years ago

Actually I was wrong about parallelization, its already supported by the wrappers.

Actually, some steps are multi-threaded (cstacks) but the other take or should take dataset collection and run the data individually in batch (ustacks, gstacks and sstacks)

On the other hand, the "wrappers scripts" (not the Galaxy one) denovomap and refmap just lauch the individual steps and aren't so efficient for the parallelisation.

bernt-matthias commented 6 years ago

It seems that also u, g, and sstacks take a threads parameter, so I guess they run parallel and in batch.

lecorguille commented 6 years ago

Indeed :)

bernt-matthias commented 6 years ago

step1: https://github.com/bioconda/bioconda-recipes/pull/10942

bernt-matthias commented 6 years ago

started also step2: https://github.com/galaxyproject/tools-iuc/pull/2085 comments welcome

bgruening commented 6 years ago

Thanks @bernt-matthias! Stacks in now available and I restarte #2085

bernt-matthias commented 6 years ago

Boah. There seem to be so many changes in stack 2.x. Some programs are gone (eg genotypes) and many parameters are new and others gone, some are marked as not implemented ... populations -P seems to expect some catalog.gz file ....

Is there some experienced user that could point me to some docs or provide some command lines + tests data that I could use as starting point.

Anyway. I guess this will result in in new wrappers ..

bernt-matthias commented 5 years ago

https://github.com/galaxyproject/tools-iuc/pull/2085