getting weird volcano plot

Question

[Shero]

Hello all,

I'm looking into the differentially expressed genes in heat-stressed animals versus normal ones.

When I only done the analysis on the 6 samples (3 heat stressed and 3 non-stressed) I got a few significant genes that are significantly differentially expressed on stress.

When I did the DEseq analysis using data matrix of all samples that include RNA-seq of 40 samples 3 of them are heat stressed the other 3 are non-stressed and the rest are of different other conditions like ageing and cold stress.

dds <- DESeqDataSetFromMatrix(countData = data,
                              colData = samples, 
                              design = ~0+condition)

and in my DEseq contrast

ageing <- results(dds_lrt,
                   contrast = c("condition", "5.months.old", "young"))

stressed <- results(dds_lrt,
                   contrast = c("condition", "heat-stressed", "control"))

by applying that I got hundreds of significant genes.

Is this a right way of doing it and the volcano blot is weird

Any recommendations?

[Hyper_Odin]

Looks fine to me. It will look better if you transform the log(padj) values to log10 or log2.

[jv]

This looks a bit off to me, but we don't know enough of the details to really interpret. Your first line of code creates object dds but then subsequent lines use object dds_lrt, can you please show your complete code? What is the difference between dds and dds_lrt? Are you trying to do a likelihood ratio test?

Please also show the code for making the volcano plot.

Also, how many samples do you have in each condition?

[Shero]

Thank you for your reply. This is the code to get dds_LTR

dds_lrt <- DESeq(dds, test="LRT", reduced= ~ 1)

It's 3 samples per each condition. My concern is that in total I have 40 samples of completely different condition. I don't compare all of them together some of them are ageing and other heat stress some cold stress. When I include all of them in the same dds_ltr and specify which i want to compare to which it gives me tons of significant DEgenes but when I split each condition and the control of it to different dds_ltr so the data will have only 4 column or 6 column it parely give me something pass padj <0.05 is that normal?

the code for the volcano plot is

 ggplot(data=stressed %>% arrange(threshold), aes(x=log2FoldChange, y=
 -log(padj))) +   geom_point( size=1.5, aes(color=threshold)) +theme_classic()+   scale_color_manual(values=c("darkgrey","darkseagreen4","darkorchid4"))+
 theme(
     axis.line.x  = element_line(colour = 'black', size = 1),
     axis.line.y  = element_line(colour = 'black', size = 1),
     legend.text = element_text(family = "Arial",
                             color = "black",
                                size = 8, face = "plain"))

[Shero]

I have 3 repeat per each condition. what is wierd is that I have alot of genes with padj <0.05 and alot of padj that is zero the code for volcanoplot is

ggplot(data=stressed %>% arrange(threshold), aes(x=log2FoldChange, y=
 -log(padj))) +   geom_point( size=1.5, aes(color=threshold)) +theme_classic()+   scale_color_manual(values=c("darkgrey","darkseagreen4","darkorchid4"))+
 theme(
     axis.line.x  = element_line(colour = 'black', size = 1),
     axis.line.y  = element_line(colour = 'black', size = 1),
     legend.text = element_text(family = "Arial",
                             color = "black",
                                size = 8, face = "plain"))

[Ram]

Shero Why did you delete this post?