Mismatch between MAST DE results and dotplot

Dear community,

I found that a gene that has only a small fold-change between two conditions based on MAST, conversely appears to have a large difference if plotted in a dotplot:

image

MAST results:

gene	log2FC	  pct.1	  pct.2
Flt1	-0.02    0.98	  0.99

The effect size is not huge, but in the dotplot the difference looks even stronger that that of the MAST top hit (which has a 50-fold smaller pvalue).

Unscaled normalized values were used for MAST.

Please note that the gene is highly expressed. Maybe that has an effect on MAST?

Any advice is appreciated!