[R-sig-ME] info model glmer
Guidantonio Malagoli Tagliazucchi
gu|d@nton|o@mt @end|ng |rom gm@||@com
Fri Mar 6 21:30:34 CET 2020
Hi,
I have a dataset with two main groups (e.g. healthy and disease, column
"status") and inside each group other two groups (status_genomic). I have
another column with the tissues (tissue). Columns are genomic features.
input_df<-data.frame(status=as.factor(sample(0:1,2000,replace=T)),
status_genomic=as.factor(sample(0:1,2000,replace=T)),
tissue=paste('tissue',sample(c('a','b','c','d'),2000,replace=T),sep='_'),
replicate(40, rnorm(2000)))
colnames(input_df)[-c(1:3)]<-paste(rep(paste('genomic_feature'),40),1:40,sep='_')
Rows are subjects, columns genomic features.
My aim is identifying of the genomic features (genomic_feature) that are
predictive of the status (column status) and status_genomic. For example I
would like to identify the genomic_feature X with a relation with the
status but also with the status_genomic. From my model I would like to
exclude the effect of the tissue that is a confounder.
I thought to the following model:
my_formula<-as.formula(status ~ genomic_feature1 + genomic_feature2 +
genomic_feature3 [...] + status_genomic + (1 | tissue))
glmer(my_formula, input_df, family= binomial("logit")
I am not totally sure that this is correct (especially for status_genomic).
Do you have any suggestion who you can provide to me to develop a better
model? Sorry for this question, but i am not an expert.
Thanks in advance for the support,
Gui
[[alternative HTML version deleted]]
More information about the R-sig-mixed-models
mailing list