This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Histogram of p values for all tests

Hi, I am trying to do differential-gene-expression analysis with DESeq2, for that i have done pseudo-alignment using Salmon , and with tximport, gene-level abundance is calculated. I am following "Analyzing RNA-seq data with DESeq2", tutorial. I got Histogram of p values for all tests.

enter image description here

Please, guide me to interpret this plot. Can I proceed with the data, (for heatmap generation), showing above plot.

Thanks in advance!

rna-seq deseq2
• 6,557 views • link
updated by tw617 • tw617 • written by shivyasoni1994 • shivyasoni1994

Correct what? Please specify.

• 1 views • link
written by ATpoint ATpoint

2 answers

Start reading this excellent article here : http://varianceexplained.org/statistics/interpreting-pvalue-histogram/

These two peaks at the right tail are strange. Did you use the nominal pvalue or the corrected one (you should use the nominal for the plot) ?

Otherwise the distribution seems to be anti conservative so it's good 😉

• 0 views • link
written by Nicolas Rosewick Nicolas Rosewick

The two main patterns you should look for when interpreting p-value histograms are: a sharp spike near 0 ( like you have here) that drops off into a somewhat uniform distribution. This indicates that you have some significant differences. The other pattern would be a totally uniform (flat) distribution, this indicates no significant difference. Here it looks like you have the first and ideal pattern although those spikes near one are a bit weird. I think they are probably errors/genes with no or little reads that were not filtered out. Usually you should exclude genes with less than 10 total reads.

• 0 views • link
written by tw617 • tw617

Log in to answer this question.

More posts like this