MAST compare two conditions with several donors
1
0
Entering edit mode
@sofiapuvogel-23835
Last seen 7 months ago

Enter the body of text here

Hello, I am using MAST to find DE through two different conditions (disease and health) in a single-nucleus transcriptomic data, and so far I have cells obtained from 3 sick subjects and 3 healthy subjects. The cells obtained from the 6 subjects were sequenciated all together therefore I have a single "batch". I am looking for DE in a single cell type and comparing between the two conditions, disease and health. How do I include the variable "donor identification" in the model? Should I add the donor id as a covariate? What happens when you have many donors, do I add each donor id as a covariate as well? Thank you very much for your help!

Code should be placed in three backticks as shown below


# include your problematic code here with any corresponding output
# please also include the results of running the following in an R session

sessionInfo( )

MAST • 167 views
0
Entering edit mode
@andrew_mcdavid-11488
Last seen 4 months ago

You can't model both donor and disease as covariates, because donor is confounded with disease. However you could apply a donor random effect, eg,

zlm(~ disease + (1| donor), method = 'glmer')


With sufficient donors, this would be the preferred model.

0
Entering edit mode

Hi Andrew, sorry for jumping into this question but I have a similar case. Regarding your answer, how can I tell which number of donors is sufficient? Also, in my case, the cells are not uniformly distributed amongst the donors. I suspect this might affect the model, is the random effect still the way to go? Any suggestion would be greatly appreciated. Thanks