How To Compute P-Values From Gene Expression Data?

Question

[Sanchita Gupta]

I would like to know, how to calculate p-values from log2 (expression ratio) value of given set of genes in a replicated microarray experiment.

[Steve Lianoglou]

Here's the entrance to one rabbit hole, enjoy: http://www.bioconductor.org/packages/2.9/bioc/vignettes/limma/inst/doc/usersguide.pdf

[Michael Kuhn]

Sheesh, don't bite the newbies. If you're new to a field, you don't know the right keywords to search for (and get decent results).

Answer 1

[Michael]

I'll treat this as a sort of FAQ question, let's see if it pays off:

p-values are the results of applying a statistical test. The essential requirement is that the experiment is replicated (at least 3 repeated measurements is mostly recommended).

For a basic overview: http://en.wikipedia.org/wiki/Gene_expression_profiling#Statistical_analysis

There are many software tools that can do these calculations (even Excel could do a simple t-test), but R and Bioconductor contain the largest collection of microarray test-tools: e.g.:

• limma
• siggenes

• RankProd

• multtest for dealing with the multiple testing problem

For differential expression analysis of RNA-sequencing, there are e.g.:

• edgeR
• DESeq
• DEGseq

• cuff-links/-diff (stand alone)

• here's the complete list.