search

Irrationone / cellassign

Automated, probabilistic assignment of cell types in scRNA-seq data
Other
191 stars 79 forks source link

Questions about parameters in cellassign function #56

Closed lxhuang7 closed 4 years ago

lxhuang7 commented 4 years ago

Hi,

Thanks for this very useful tool! I would like to ask two questions about cellassign() function.

(1) I'm a bit confused on the default value of the parameter min_delta = 2. In the documentation it is described as the minimum log fold change a marker gene must be over expressed. Is this 2 the fold-change or log2-fold change? I'm asking because we often use 2-fold change (1-log-fold change) to describe the expression changes. If it's log2-fold change, the actual fold-change would become 4.

(2) I noticed you use scran::computesSumFactors() to compute the library size factors. Do you happen to know how it affects cell assignment if scater::librarySIzeFactors() is used instead?

Thanks! Lei

Irrationone commented 4 years ago

Hi Lei,

Good questions:

1) min_delta corresponds to the log fold change here. So yes, a min_delta of 2 would correspond to a minimum fold change of exp(2).

2) In practice, this shouldn't make a substantial difference in terms of cell assignments. The core assumption behind CellAssign is only that a marker gene for a given cell type is over-expressed to around the same degree in cells of that type. The only time you might run into issues is if you're dealing with cells that have very low size factors relative to others, but these should largely be filtered out in whatever QC steps you have in your workflow.

Allen

lxhuang7 commented 4 years ago

That makes sense. Thanks Allen!