Repository navigation
AleRax - #8900
AleRax#8900aparamitarana009-boop wants to merge 6 commits into
Conversation
github-project-automation
Bot
moved this to Triage/Discuss
in Galaxy Tool Developers SIG Board
|
please review |
…elp, add co-author
| NP=\${GALAXY_SLOTS:-1} && | ||
| if [ "\$NP" -gt $n_families ]; then NP=$n_families; fi && | ||
| if [ "\$NP" -gt 1 ]; then | ||
| MPI_PREFIX="mpiexec --allow-run-as-root --oversubscribe -np \$NP"; | ||
| else | ||
| MPI_PREFIX=""; | ||
| fi && |
There was a problem hiding this comment.
Capping NP at the number of families is a nice touch. One thing though: CI runs with GALAXY_SLOTS=1, so the mpiexec branch never actually runs in the tests. Did you try it locally with something like GALAXY_SLOTS=2 planemo test? The bioconda package uses OpenMPI, so --allow-run-as-root --oversubscribe should be fine, but it'd be good to know it really works in the container. Could you confirm you've run it with more than one slot?
There was a problem hiding this comment.
I tried it with GALAXY_SLOTS=2 on my Mac, but I have an Apple chip, and the AleRax biocontainer is only built for linux/amd64 (bioconda has no ARM build of alerax). So Docker has to emulate it, and it already fails at MPI_Init, even with just one slot.
There was a problem hiding this comment.
CI should run with more than one slot. I remember 2 / 4? Can you output some test string to see if this is happening. Can we see somehow else if it is working.
@aparamitarana009-boop does bioconda support ARM in general, then bumping the build over there might help, i.e rebuild the package.
| </configfiles> | ||
| <inputs> | ||
|
|
||
| <conditional name="species_tree_input"> |
FOR CONTRIBUTOR:
This PR adds a wrapper for AleRax (v1.4.0, available on Bioconda), a tool for species tree inference and gene tree reconciliation under gene duplication, transfer and loss.
Inputs: a species tree (or a MiniNJ / random starting tree), a list collection of gene tree samples (one element per gene family), and an optional list collection of gene-to-species mapping files.
Options: species tree search (SKIP / REROOT / HYBRID), reconciliation model (UndatedDTL / UndatedDL), model parametrization (GLOBAL / PER-FAMILY / PER-SPECIES), number of samples and random seed.
Outputs: species tree, log, per-family likelihoods, per-family event counts and transfers (as collections), and an optional .tar archive of the full output folder.
The test data is a small subset (two families) of the primates example from the AleRax repository. All 7 tests pass locally with
planemo test, andplanemo lint/planemo shed_lintare clean.This is my first tool for IUC, so any feedback is very welcome.
@Deeptivarshney could you have a look for review? Thanks!