do not list negative contributors for NMF projection or any gene cart with no negative weights

Question

do not list negative contributors for NMF projection or any gene cart with no negative weights

Closed this issue a year ago · 11 comments

carlocolantuoni commented a year ago

there are lots of zeros in the NMF weights - there are not of interest as the low weights in PCA are

adkinsrs commented a year ago

Committed

Answer 1 · 2023-09-13T19:06:02.000Z

Just hide any cells where the weight is 0 or below?

Answer 2 · 2023-09-13T19:26:14.000Z

This is for the text you display at the top of the projection page, highlighting genes with high or low weights - not for the display of cells. Does it make sense with that in mind?

…

On Wed, Sep 13, 2023, 15:06 Shaun Adkins ***@***.***> wrote: Just hide any cells where the weight is 0 or below? — Reply to this email directly, view it on GitHub <#593 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH7KC7VAGY6LJHKFCBQCBTTX2H72JANCNFSM6AAAAAA4NJMDTA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

Answer 3 · 2023-09-13T19:27:14.000Z

Ahh, yes... Just hide the top 5 "low weight' section.

Answer 4 · 2023-09-13T19:29:34.000Z

Yup, do this for NMF or if the weights contain no negative values i.e. if 0 or above is max of the weights

…

On Wed, Sep 13, 2023, 15:27 Shaun Adkins ***@***.***> wrote: Ahh, yes... Just hide the "low weight' section. — Reply to this email directly, view it on GitHub <#593 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH7KC7XCXEWEDHH42NUUXYTX2ICJ3ANCNFSM6AAAAAA4NJMDTA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

Answer 5 · 2023-09-14T17:19:34.000Z

@carlocolantuoni , what should we do if in a given pattern source file, pattern 1 has negative weights, and pattern 2 does not have negative weights? Do we only show them for pattern 1, or do we show for all patterns?

Answer 6 · 2023-09-14T17:57:53.000Z

for any NMF we can just disable all reporting of the "negative contributors". for other gene carts, i think its fine to report pattern by pattern - i.e. list the individual genes under "negative contributors" if there are negative values in the pattern, and report "no negative weights" instead of gene names if there are no negative values for that particular pattern.

…

On Thu, Sep 14, 2023 at 1:19 PM Shaun Adkins ***@***.***> wrote: @carlocolantuoni <https://github.com/carlocolantuoni> , what should we do if in a given pattern source file, pattern 1 has negative weights, and pattern 2 does not have negative weights? Do we only show them for pattern 1, or do we show for all patterns? — Reply to this email directly, view it on GitHub <#593 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH7KC7W744HLPAGS3KEIXEDX2M4DBANCNFSM6AAAAAA4NJMDTA> . You are receiving this because you were mentioned.Message ID: ***@***.***>

-- Carlo

Answer 7 · 2023-11-02T17:05:24.000Z

hey @adkinsrs - im still seeing the "Strongest negative contributors" for NMF projections - is this another element that will go in later with the big overhaul?

Answer 8 · 2023-11-02T17:10:21.000Z

The commit was added to the main branch, not the v2 work. I believe it's just a case where the latest code has not been pushed to the nemo production server yet.

Answer 9 · 2023-11-02T17:12:07.000Z

@jorvis has some commits after my Sept 14th one, so I want to defer to him if those commits are safe to push to production

Answer 10 · 2023-11-04T06:49:11.000Z

I don't have any commits on main which shouldn't be tested for production