stephaniehicks/qsmooth

1 group?

Closed this issue · 3 comments

OnlyBelter commented

I just read your wonderful paper and would like to ask a question.

What's the meaning of the results if I assign 1 single group for all samples? Is qsmooth equivalent to standard quantile nomalization under this situation?

The problem I want to solve is that simulated bulk RNA-seq data by single cell data (dataset1) separated far away from real bulk RNA-seq data (dataset2, such as data from TCGA) if we check the first two principal components. Similar to the third column of Fig. 5., dataset1 and dataset2 are two separated groups in PC1/PC2 figure. We want to eliminate the difference between two groups. Can we say this difference is caused by batch effect and can be removed by quantile normalization?

Thanks!

stephaniehicks commented

Hi @OnlyBelter, yes in the case that there is only 1 group, I believe it is the same as standard quantile normalization.

perinom commented

Just to follow up on the question, currently if there's only 1 group a standard qsmooth() call fails with

Error in `contrasts<-`(`*tmp*`, value = contr.funs[1 + isOF[nn]]) : 
  contrasts can be applied only to factors with 2 or more levels

Is there a way to run standard quantile normalization from within qsmooth in these cases or should I use external resources?

stephaniehicks commented

Hi,

If there is only one group, you are essentially doing quantile normalization. qsmooth was designed for at minimum two or more groups. Thanks!