Estimated FDP plot - interpretation

[MiriamAng]

Hi,
I am using subSeq to evaluate the effect of sequencing depth on a 3' mRNA-seq experiment sequenced at 20M reads.
As suggested in the package vignette, I simulated four additional sequencing depths by picking subsamplings on a logarithmic scale as follows:
proportions = 10**seq(-2, 0, 0.5)
I then compared two differential expression analysis methods, i.e. DESeq2 and limma, by creating customized functions that accept the desired contrast.

Among the 4 plots obtained, 3 looked quite fine to me (or at least somehow reflecting the examples shown both in the vignette and manuscript). However, when coming to the estimated FDP plot, I got a trend that I am not able to interpret (please find the pictures below)

1st contrast

2nd contrast

(Note! The MSE for myDEseq2 is not flat, it seems so due to the huge difference in scales)

From the package vignette it is stated the following:

In light of this, what conclusions can be derived from my plots? I would be very grateful if you could help me in correctly interpreting the above plots. If need, I can also provide the R script.

Thank you for your time.

Miriam