Question

Why do I have this perfect volcano plot from Limma?

0

Entering edit mode

SmallChess ▴ 10

@smallchess-7765

Last seen 16 months ago

Australia

I have a question on limma differential testing on proteins.

https://bioinformatics.stackexchange.com/questions/4850/why-do-i-have-this-perfect-volcano-plot-from-limma

I have the script that can reproduce the plot in the link. My question is why Limma would generate the volcano plot as described, as I failed to see anything suspicious in my data set.

limma • 3.3k views

ADD COMMENT • link updated 6.3 years ago by Gordon Smyth 51k • written 6.3 years ago by SmallChess ▴ 10

0

Entering edit mode

chris86 ▴ 420

@chris86-8408

Last seen 4.9 years ago

UCL, United Kingdom

I can only assume this is something weird about the data you have acquired as the analysis code looks fine to me at least.

ADD COMMENT • link 6.3 years ago chris86 ▴ 420

score 5 · Accepted Answer · 2018-08-07

5

Entering edit mode

Gordon Smyth 51k

@gordon-smyth

Last seen 8 hours ago

WEHI, Melbourne, Australia

It means that your data has df.prior = Inf so that all the genes share the same pooled variance. Hence the p-values are a monotonic function of fold change.

ADD COMMENT • link 6.3 years ago Gordon Smyth 51k

0

Entering edit mode

Thanks for your response. Is this a problem in statistical inference? Is there anything I should do to minimize the effects?

ADD REPLY • link 6.3 years ago SmallChess ▴ 10

0

Entering edit mode

Whether or not this is symptomatic of a problem depends on what the data is, which is something you don't give much information about.

If this is real proteomic data from biological replicates, then this phenomenon is not a good sign. It can often arise from a strong batch effect that hasn't been accounted for.

If you want to discuss this, then please post a new question with some relevant background about your experiment. But first explore your data by making some plots (plotMDS, plotMD, plotSA etc).

ADD REPLY • link 6.3 years ago Gordon Smyth 51k