I've been working with DEseq2 for a while now, and have been going about plotting the normalized counts for a couple genes I found.

I'm following the vignette subsection "Plot counts", to plot the counts using ggplot2, but I noticed the authors used a specific set of values for scaling their y axis (25,100,400).

I'm wondering if there is a best practice for choosing the scale of the y-axis here, or if these values were chosen based on how DEseq calculates normalized counts as well? Including link to specific section on the vignette below as well.

https://www.bioconductor.org/packages/release/bioc/vignettes/DESeq2/inst/doc/DESeq2.html#plot-counts

genex <- "ENSMPUG00000004881"
x <- plotCounts(dds, gene=genex, intgroup="dpi", main = "ENSMPUG00000004881", returnData = TRUE)
y <- plotCounts(dds, gene=genex, intgroup="severity", main = "ENSMPUG00000004881", returnData = TRUE)
x$severity <- y$severity

plot <- ggplot(x, aes(x=dpi, y=count, color=severity)) + 
  geom_point(position=position_jitterdodge(jitter.width=0.25), size=5) + 
  scale_color_manual(values = c("darkgreen", "gold", "darkred")) +
  scale_y_log10(breaks=c(25,100,400)) + # Why 25,100,400 ???
  geom_smooth(aes(group=severity),method="loess",se=TRUE,fullrange=TRUE) +
  ggtitle("ACCS3")+
  xlab("Day(s) Post Infection")+
  ylab("DESEQ2 Normalized Counts")+
  labs(color="Severity")+
  theme_bw()
sessionInfo( )

It's just visualization. Don't overthink it and choose what you feel is appropriate to send the message for your analysis. I doubt that there is a gold standard answer for this other than "it looks good". Generally, the only rule I know of is that you should not mess with axis limits to mislead the reader, for example making limits very small so that tiny differences look big.