volcano plots cummeRbund

[godzilla07]

I am using the volcano plots to display significantly expressed genes but I see a cutoff at -log10(Pvalue) to be around 2.0 which is a pvalue of 0.01. I should be seeing the cutoff at around 1.3 for pvalues < 0.05, which according to the cummeRbund paper is the default cutoff. How can I adjust the plots to reflect the real cutoff? I am not familiar with R or ggplot.

[glados]

Try
csVolcano(genes(cuff), 'x', 'y', alpha=0.05, showSignificant=T)

[jmwhitha]

Followup

This is very useful. Thank you!

I'd like to ask a similar question about these plots. Do either of you (or anyone else) know how to change the p-value limit? I have a ceiling of significant differential expression points on my volcano graph. Seems to me that cuffdiff only tested or reported p-values up to 0.0001 since -log(0.0001) = 4. I actually think the cut off is at 0.001 and they add 1 (maybe more) to the -log(p-value) because when I use alpha=0.0005 or alpha=0.0001 in the command, nothing is significant. Please see attached graph to see what I mean by "ceiling".

Also, I was wondering if you know how to determine which genes or CDS correspond to the points on the volcano or scatter plot.

Thanks and God bless,
Jason

[AJenkins]

Hi Jason,

Did you ever find an answer to this question? I am having a similar problem with my volcano plots with a "ceiling" of the same as yours and am wondering how to correct for it?

-Adam

[jmwhitha]

Hello Adam,

I actually ended up using a simpler pipeline. I found too many mysteries with TopHat-Cufflinks...Cummerbund, the kind that you need faith for. Though I have faith in my Savior, to me, there's too many pipeline options to put my faith in this one.

To better answer you question, I did not find an option for changing the p-value. I am sure that the option would need added to a command before Cummerbund though since all Cummerbund is doing is displaying the coordinates given to it. What I was going to do before I decided to switch pipelines is adjust the x axis so it looks better and make a note in the figure. Something like, "p-values less than p = 0.0001 were rounded to 0.0001; therefore, some genes' differential expressions are actually more significant (p < 0.0001).

Here is a few commands I found useful for generating an image and adjusting the x axis. To be clear, even if you could adjust the y-axis, it would not change the coordinates of the "ceiling" of differential expression expression values.

png(file="csVolcano_genes1.png")
csVolcano(genes(cuff_data_1), 'sample1', 'sample2', alpha=0.05, showSignificant=T, xlimits = c(-10, 10))
dev.off()

You can switch "genes" with "CDS" or "isoforms" if you want.

Peace and God bless,
Jason

[AJenkins]

Hi Jason,

I just found the reasoning for why cuffdiff doesn't report any p-value less than 0.0001, and it can be found at this website: http://sihua.us/Cufflinks.htm

In short it says: The range of p-values that users should expect from Cuffdiff has changed. Because the test is now based on explicit sampling from the beta negative binomial, users will not see values less than 10^-5 by default. The test_stat field of Cuffdiff's output still contains the delta method's test statistic, but this test statistic is not used to compute p-values. It is preserved for backward compatibility with some functions in CummeRbund.

Therefore we are seeing a ceiling from which is causing by this minimum p-value it will report.

Hope all is well,

Adam

[jrb71]

Hi All,

I have a similar question as to the one that started this thread. I've used this expression to produce a volcano plot:

>csVolcano(genes(cuff_data), 'sample1', 'sample2', alpha=0.05, showSignificant=T, xlimits=c(-10,10))

I was wondering if it is possible to include an argument to specify the log2_fold_change as well as the alpha value for colouring the data points.

Thanks
Jen

[AJenkins]

Hi Jen,

I'm not sure if there is a way to use the log2foldchange for coloring of the data points. You'd definitely have to open it up to see how they define the variables. You may be better off emailing Loyal Goff (the creator) to see if there is an easy way. He's usually very good at responding to emails/questions (i've found a couple bugs previously).

-Adam

[AndrewO]

Hi All,

I have a similar question as to the one that started this thread. I've used this expression to produce a volcano plot:

>csVolcano(genes(cuff_data), 'sample1', 'sample2', alpha=0.05, showSignificant=T, xlimits=c(-10,10))

I was wondering if it is possible to include an argument to specify the log2_fold_change as well as the alpha value for colouring the data points.

Thanks
Jen

Hi Jen,

This might be beneath you but might be helpful nonetheless for others.

We have an application, called iPathwayGuide that is free to use and can take a CuffDiff file (among others) and allow you to specify the log2 FC and p-value cutoff and provide a volcano plot with either just the DEGs, or see all the genes with the DEGs highlighted.

It would look something like this: