@LTLA is a pretty awesome dude!
He suggested: Use the log2FC in expression between the empty drops of two conditions as NULL Hypothesis. Test the log2FC obtained from comparisons of the actual groups against the NULL log2FC.
I now removed genes with logFC > 2 between the ambient profiles and performed DE on the remaining genes (differences in ambient profiles only occur when spermatids are present in one condition but not the other)
@LTLA is a pretty awesome dude! He suggested: Use the log2FC in expression between the empty drops of two conditions as NULL Hypothesis. Test the log2FC obtained from comparisons of the actual groups against the NULL log2FC.