usher icon indicating copy to clipboard operation
usher copied to clipboard
verified publisher icon yatisht

ENH: Display all maximally parsimonious locations in web interface, not just 1

Open corneliusroemer opened this issue 3 years ago • 4 comments

When I tried to place two sequences they showed 4/8 maximally parsimonious placements. However, there seems to be only 1 subtree for each with 1 placement.

How is this placement chosen? If they are all equally parsimonious, it sounds like that other placements would be just as good. Is it possible to get subtrees for those other placements too? That'd be really neat. Either one subtree per placement, or maybe even cooler, one big subtree with all the possible placements. If there are 5 placements, each would get it's mini-neighbourhood of say 100 nearby sequences, the other 500 would be global tree subsampled.

corneliusroemer avatar Feb 07 '22 13:02 corneliusroemer

Hi @corneliusroemer, have you tried the --multiple-placements option in command line UShER?

yatisht avatar Feb 07 '22 21:02 yatisht

I haven't! I was thinking this could exist in command line, but didn't look for it.

I prefer the web interface due to convenience, to quickly drop a sample. I guess my feature request is to add that --multiple-placements option to the web version.

Totally understand if this is not a priority, if this is a niche use case command line should work fine.

corneliusroemer avatar Feb 08 '22 05:02 corneliusroemer

Adding @AngieHinrichs to see if she has any thoughts on adding this option to the web interface.

yatisht avatar Feb 08 '22 17:02 yatisht

Don't worry if this is difficult. I just feel it could add a lot of value. If all the possible placements are very closely together it means something very different than if they are scattered through the global tree.

How do you currently choose which of the most parsimonious placements to display on web? Is it random?

On Tue, Feb 8, 2022, 18:12 Yatish Turakhia @.***> wrote:

Adding @AngieHinrichs https://github.com/AngieHinrichs to see if she has any thoughts on adding this option to the web interface.

— Reply to this email directly, view it on GitHub https://github.com/yatisht/usher/issues/210#issuecomment-1032858355, or unsubscribe https://github.com/notifications/unsubscribe-auth/AF77AQPUNE4Q4OKG3PPRE43U2FFI3ANCNFSM5NXQUP2Q . You are receiving this because you were mentioned.Message ID: @.***>

corneliusroemer avatar Feb 08 '22 17:02 corneliusroemer