search

mohuangx / SAVER

Single-cell RNA-seq Gene Expression Recovery
110 stars 18 forks source link

is it possible to do correlation analyses without running the correction #35

Open hk20013106 opened 2 years ago

hk20013106 commented 2 years ago

I have run the tutorial successfully and I wonder is it possible to run the correlation because I have already done imputation using another software. For example, does it make sense to generate the SE matrix manually using only values of zero?

Thanks!

Kai

mohuangx commented 2 years ago

Hi Kai,

If the SE matrix is all zero, then the cor.genes and cor.cells functions will be identical to taking cor(t(x)) and cor(x) for expression matrix x. If you want to estimate the uncertainty of the imputation from the other software, you can set mu in the saver function to the imputed values (see "Using other predictions" of the tutorial). For example, out <- saver(x, mu = other_impute) and then run cor.genes or cor.cells on out.