I performed the statistical analysis on a merged dataset.
This merged dataset was preprocessed as follows:
I had one dataset containing stromal cells and one dataset containing T cells (both Seurat objects).
I merged the countmatrices and performed SCTransform.
The outcomes of this normalization step were used as imput of CellphoneDB.
What I observed are a lot of interaction pairs with means below 1 (but above zero).
However, in the dotplot (where values are log2 transformed) these obviously become negative.
Are these findings then still biologically relevant? Since I do find differences between clustercouples than would fit my hypotheses, but I'm puzzled by the low mean + negative log2 result.
Hi guys,
I performed the statistical analysis on a merged dataset. This merged dataset was preprocessed as follows: I had one dataset containing stromal cells and one dataset containing T cells (both Seurat objects). I merged the countmatrices and performed SCTransform. The outcomes of this normalization step were used as imput of CellphoneDB.
What I observed are a lot of interaction pairs with means below 1 (but above zero). However, in the dotplot (where values are log2 transformed) these obviously become negative.
Are these findings then still biologically relevant? Since I do find differences between clustercouples than would fit my hypotheses, but I'm puzzled by the low mean + negative log2 result.
Best, Madelon