search

statOmics / tradeSeq

TRAjectory-based Differential Expression analysis for SEQuencing data
Other
237 stars 27 forks source link

Adapting to cytometry data? #221

Open 83years opened 1 year ago

83years commented 1 year ago

Hi,

I am trying to apply a pseudotime analysis to flow cytometry data. These data a processed through a pipeline similar to the CytofWorkflow (Nowicka et al., 2017) and we generate sce objects that can be run through the Slongshopt package (Risso et al., 2018). However, when it comes to running the tradeSeq package I can't get past the issue that we use expression values (that can be negative) not gene counts in our data.

Are there any work-arounds that can get me on the right path?

koenvandenberge commented 1 year ago

Hi @83years ,

I have limited experience with flow cytometry data, but I know that a biexponential transformation is quite common and indeed leads to negative and positive transformed values. The distributions of the transformed MFIs can have funny shapes, but if you are willing to assume a parametric distribution, you can change this using the family argument in fitGAM by e.g. setting it to "gaussian".

I would be careful with interpreting the results of the statistical inference, though.