Question: unequal sample sizes in groups for testCovariate argument in dmrseq
0
15 days ago by
shicks30
USA/Baltimore/Johns Hopkins
shicks30 wrote:

Hi,

I have N=44 WGBS samples from purified cell types from whole blood samples. I'm interested in identifying DMRs for six purified cell types (DMR in Bcell vs else, CD4T vs else, etc), but I have unequal sample sizes in each group:

> table(pheno_table\$cell_type)

Bcell  CD4T  CD8T  Gran  Mono    NK
4    12     4    14     6     4


My two questions are:

(1) Will the unequal sample sizes be a problem for dmrseq?

(2) My understanding based on reading the dmrseq vignette is that I can give testCovariate a multi-level factor variable with 6 levels. Is it possible to give any other design matrix?

Thank you! Stephanie

dmrseq • 39 views
written 15 days ago by shicks30

Hi Stephanie,

(1) First, the unequal sample sizes per group are not a problem; dmrseq does not assume equal sample sizes in each level. You will need to filter out any sites that are not covered in at least two samples per group however.

(2) Yes, you are correct. You can pas in a multilevel factor variable as the testCovariate. You may also add in covariate(s) you wish to adjust for in adjustCovariate.

Hope that helps, and please let me know if you have any further questions.