[BioC] Limma: Top 10% gene list based on p-values

[Ovokeraye Achinike-Oduaran]

Thanks Sean.

Duly noted. I understand that the unadjusted p-values are not
corrected for multiple testing. But thanks a bunch.

-Avoks

On Wed, May 9, 2012 at 2:21 PM, Ovokeraye Achinike-Oduaran
<ovokeraye at gmail.com> wrote:
> Hi all,
>
> Thanks to Dan, I now know how to get genes highlighted in a Volcano
> plot based on p-values from the results of my limma analysis. I would
> like to get a list of the genes highlighted. For example, the code
> below gives me a volcano plot with the top 10% genes based on p-values
> highlighted.
>
> windows()
> pdf("VolcanoPlot_GSExxxxxx_10perc.pdf");
> results$threshold = as.factor(results$P.Value<=quantile(results$P.Value, 0.1))
> g = ggplot(data=results, aes(x=logFC, y=-log10(P.Value), colour=threshold)) +
>  geom_point(alpha=0.4, size=1.75) +
>  opts(legend.position = "none") +
>  xlim(c(-5, 5)) + ylim(c(0, 8)) +
>  xlab("log2 fold change") + ylab("-log10 p-value")
> g
> dev.off()
>
> This code below gives me the exact 10% threshold value:
>
> results = topTable(fit2, adjust ="BH", number =
> nrow(gse25724dat[[1]]), sort.by="P")
> results2 = quantile(results$P.Value, 0.1)
>
> What I would like however is a list/table of the genes that meet this
> criteria, i.e, the genes highlighted from the first code. Any help
> will be greatly appreciated.
>
> Thanks.
>
> -Avoks