search

const-ae / glmGamPoi

Fit Gamma-Poisson Generalized Linear Models Reliably
105 stars 15 forks source link

compare multiple pairs concurrently #40

Open pchiang5 opened 2 years ago

pchiang5 commented 2 years ago

Hi,

My dataset contains single-cell data with the following 6 treatments:

A A+B A+B+C A+C A+C+D A+B+C+D

I would like to know the genes specifically induced by B. I wonder if it is okay and more robust to perform pairwise comparisons as below and get the aggregated output of significant genes induced by B in all conditions? A vs A+B A+C vs A+C+B A+C+D vs A+C+D+B

Thanks!

const-ae commented 2 years ago

Hi,

I wonder if it is okay and more robust [...]

'More' compared with what?

A vs A+B A+C vs A+C+B A+C+D vs A+C+D+B

All these three comparisons are equal to just testing for the effect of B (assuming you have a one-hot encoding for your conditions).

Best, Constantin

pchiang5 commented 2 years ago

Hi Constantin,

My concern is including C in the 2nd group and C&D in the 3rd group might lead to the presence of different sets of differential expressing genes. By pseudobulking the 3 pairs as separate conditions, is it possible to get a more consistent (robust, regardless the confounding 'irrelevant' factors) ones responding to +B?

Best, Po

On Tuesday, November 15, 2022, Constantin @.***> wrote:

Hi,

I wonder if it is okay and more robust [...]

'More' compared with what?

A vs A+B A+C vs A+C+B A+C+D vs A+C+D+B

All these three comparisons are equal to just testing for the effect of B (assuming you have a one-hot encoding for your conditions).

Best, Constantin

— Reply to this email directly, view it on GitHub https://github.com/const-ae/glmGamPoi/issues/40#issuecomment-1315067877, or unsubscribe https://github.com/notifications/unsubscribe-auth/AG6RZ3KLHJOUSELOLQQRPLDWINNE5ANCNFSM6AAAAAASAVRBSU . You are receiving this because you authored the thread.Message ID: @.***>