cummeRbund scatterplot labels

[orthodoc]

Hi Everyone,

Disclaimer: I am very new to next gen sequencing / the Tuxedo suite of tools so please excuse gross incompetence!

I am using cummeRbund to visualize the results of my Cuffdiff run and while it is very impressive, I have a few questions about the figures it generates. I have 3 replicates of 5 conditions, each condition was compared head-to-head in 4 individual cuffdiff runs (this was done by our university's bioinformatics core).

I would like to take a selection of the most significant deferentially expressed genes and present it on a scatterplot. My workflow to accomplish this has been to look through gene_exp.diff in Excel, manually filter and sort as necessary, then pick my genes of interest and go back to R to create a gene set with these id's, then plot away. It works fine up to this point.

I can create the scatterplot, but it is of limited utility unlabeled. I would like to label each of the points (I'm only using about 20 genes in each set) with what gene it represents. Is this possible?

This would make my analysis much much easier than manually looking through the gene_exp.diff tables.

Please let me know if anyone has any suggestions or if I'm approaching this in the wrong way.

Thanks!

-orthodoc

[cbaudo]

Hello orthodoc,

I was wondering if you found a solution to this problem.

Thank you,
cbaudo

[sindrle]

I recommend the CummeRbund manuals section about "Creating Gene Set".

You really don't need Excel for anything.

Going this route you may then use ggplot2 commands on your gene set and alter your plot as you like, works like a charm.

[cbaudo]

Thanks sindrle,
I've successfully imported by data into R and labeled the points on the scatterplot with gene IDs, but I am encountering another issue. I would like to modify the axis intervals in the same way cummeRbund does. Currently, a majority of my data points are clustered together with low FPKM values, but a few genes have large FPKM values which skew the graph to have larger axis. All of the points are then smooshed together.

Do you know of an R function I can use or of a work around for this? Also, do you know if cummeRbund log transforms the data prior to plotting?

cheers,
cbaudo

[sindrle]

I would read about ggplot2, its this package used by cummerbund. I can handle everything like this.

see the help() function for the plot, it should be an option called log = TRUE, or something like that.

Good luck!