MacPorts fails to install sativa directory in bin

[mcottrell]

The new sativa directory that appears in the bin directory is causing problems for MacPorts?. Would it be possible to simply place all the sativa python stuff directly in the bin directory? That would make MacPorts? a lot easier.

[westram]

Replying to mcottrell:

The new sativa directory that appears in the bin directory is causing problems for MacPorts?.

What kind of problems?

[mcottrell]

MacPorts? halts with an error when it tries to move the sativa folder into /opt/local/bin where MacPorts? keeps all the binaries. In MacPorts? a temporary workspace is used to build everything and the final step is moving things into their proper place under the MacPorts? $prefix (defaults is /opt).

[akozlov]

Replying to mcottrell:

Would it be possible to simply place all the sativa python stuff directly in the bin directory? That would make MacPorts? a lot easier.

It would create quite a mess, since there is also an underlying directory structure in sativa (e.g. a bunch of python modules under sativa/epac). In my view, the best solution would be to move the whole thing out of bin and then just symlink the top-level scripts to there - like you did for some SH and Perl scripts.

[westram]

Replying to akozlov:

In my view, the best solution would be to move the whole thing out of bin and then just symlink the top-level scripts to there - like you did for some SH and Perl scripts.

I would prefer that as well. There was already a patch necessary to handle sativa installation: [13065]

One simple way would be to move the sativa-code (that is needed in distribution) into $ARBHOME/lib/sativa (directly inside SVN), instead of maintaining it below $ARBHOME/GDE/SATIVA and then copying it around.

[akozlov]

I like $ARBHOME/lib/sativa as a final location, but do you think copying in there could also cause problems? Since it'd be nice to keep the original installation together in $ARBHOME/GDE/SATIVA and then scatter the files to corresponding locations during ARB build (apart of python scripts, there are also menu file, RAxML binaries, perl wrapper etc.). This approach seems to work for other tools, e.g. MAFFT.

[westram]

Replying to akozlov:

I like $ARBHOME/lib/sativa as a final location, but do you think copying in there could also cause problems?

No. Just handle it as you prefer

[akozlov]

OK thanks, then I'll fix it like this and commit to trunk.

[akozlov]

This should be now fixed in trunk with [13203]. Could you please test?

[mcottrell]

Good work. arb trunk/13203 now builds and runs in MacPorts?.

Now I'm curious, what new functionality does sativa add to arb?

Thanks, Matt

[akozlov]

Great!

SATIVA stands for Semi-Automatic Taxonomy Improvement and Validation Algorithm. In short, it employs RAxML to find incongruence between the taxonomy and the phylogenetic , and suggests sequences/taxa with putatively wrong taxonomic annotations.

We're now preparing a manuscript that describes this approach (joined work with Pelin Yilmaz from MPI Bremen).

Alexey

[westram]

mark changes that got fixed after arb 6.0.x

[westram]

Milestone renamed