Wish list : ILR transform from bp trees

[mortonjt]

It would be nice if there was a function in this package that could directly generate an ILR basis from a bp tree.
Basically, it would recoding the gneiss.balance.sparse_balance_basis. We can technically perform the ILR transform on 100K dimensions, but reading / creating these bases becomes a bottleneck. And this definitely won't scale well for more than 100K dimensions.

[wasade]

That would be super cool. The algorithm doesn't look very complex assuming the API methods are there but we may need to implement a method to enforce bifurcation. And i _think_ we have level order but we may not.

…

On Thu, Mar 16, 2023, 10:26 Jamie Morton ***@***.***> wrote: It would be nice if there was a function in this package that could directly generate an ILR basis from a bp tree. Basically, it would recoding the gneiss.balance.sparse_balance_basis <https://github.com/biocore/gneiss/blob/master/gneiss/balances.py#L170>. We can technically perform the ILR transform on 100K dimensions, but reading / creating these bases becomes a bottleneck. — Reply to this email directly, view it on GitHub <#53>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AADTZMX7P3U274MOKASDCO3W4NENVANCNFSM6AAAAAAV5QP5ZQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[khovvannak23]

https://github.com/khovvannak23/verbose-octo-waddle.git

[khovvannak23]

$ bp placement --help
Usage: bp placement [OPTIONS]

Options:
--placements PATH jplace formatted data [required]
--output PATH Where to write the resulting newick
[required]

--method [fully-resolved|multifurcating]
Whether to fully resolve or multifurcate
[required]

--help Show this message and exit.