Repository navigation
Discussion: future of the gene tree rooting methods (mad, Nevers_rooting, outlier) #120
Description
fastoma-infer-subhogs --gene-rooting-method accepts five values, but only the midpoint ones work. This issue is for deciding what to do with the others. They are not exposed in FastOMA.nf, so they can only be reached by calling fastoma-infer-subhogs directly.
| method | state |
|---|---|
midpoint (default) |
works |
midpoint-dendropy |
works (tested), not listed in --help |
mad |
broken: infer_rooted_genetree passes the tree object, _wrappers.mad_rooting expects a file path (_infer_subhog.py:512 vs _wrappers.py:120); the mad binary is not in the Docker image or conda env; with several equally good roots it takes the first one |
Nevers_rooting |
reads species_tree.nwk from the working directory (_infer_subhog.py:507), which doesn't exist in a pipeline task (there it is species_tree_checked.nwk); tries every edge as root with set_outgroup, i.e. quadratic |
outlier |
marked "not yet implemented completely"; debug prints (_utils_subhog.py:576-648); quadratic get_distance over all leaf pairs |
There is also an unused copy of the rooting dispatch in _utils_subhog.genetree_sd (:189-248, only called from the disconnected fragment code).
Questions
- Which methods should FastOMA offer? E.g. keep midpoint (+ dendropy variant), make
madwork (needs the binary as a dependency), and develop the species-tree-aware rooting (Nevers) properly with the checked species tree passed in. - Should the others be removed until they are ready? Removing them also reduces the ete3 surface for the ete4 migration.
- Should the option get
choices=so typos fail at argument parsing instead of at the first gene tree?
Test
tests/test_gene_tree_rooting.py::test_mad_rooting_gets_a_file(strict xfail on PR Test suite ahead of the ete3 → ete4 migration #112)
Activity