Hi all, I'm right now working with a RNA-seq raw count data file (in .txt format). It's a matrix of ~50,000 genes (row) and 8 samples (column).

My assignment is to get differential gene expression analysis. But I'm not really sure what it means by that. I'm using DESeq2 package in R for this task, and since I'm really new to R and bioinformatics, I'm not really sure what they want as the result.

So my questions are:

1) What does one expect from differential expression analysis? I've got the list of genes that have p-values less than 0.05. Would that be sufficient for differential expression analysis? What I have right now is just the list of genes.

2) What is MA plot and what does it have to do with differential expression analysis? I've read some papers about RNA-Seq and R packages that help analysis, and most of them mentioned MA plot. What is its significance in DE analysis? And what can we see from it?

Here is part of my source code that I wrote to get the list of genes that have p-values less than 0.05:

#get differential expression results
res<-results(dds)

#sort by p-value
res<-res[order(res$pvalue),]
head(res)

#write a .csv file that contains information of genes
#that have p-value smaller than 0.05
write.csv(res[which(res$pvalue<0.05),], file="DEoutput.csv")

Thanks!

MA plots were originally an invention of Terry Speed's group at UC Berkeley for two-color microarrays. MA plots first appeared in the sma and affy packages back in 2002, and were a way to display the raw data from a two-color microarray or for a pair of Affymetrix microarrays. The idea of using MA plots to display the results of a differential expression analysis appeared in the limma package in 2004.

A short review of MA plots can be found in the section on "mean-difference plots" in this article:

http://nar.oxfordjournals.org/content/43/7/e47.full

"1) What does one expect from differential expression analysis? I've got the list of genes that have p-values less than 0.05. Would that be sufficient for differential expression analysis? What I have right now is just the list of genes."

This is one possible "end point" for DE analysis. The list of genes with adjusted p-value less than X should satisfy the property that in expectation the false discovery rate in this set is less than X.

"2) What is MA plot and what does it have to do with differential expression analysis? I've read some papers about RNA-Seq and R packages that help analysis, and most of them mentioned MA plot. What is its significance in DE analysis? And what can we see from it?"

Check Gordon Smyth's pointer to a paper describing the plot.

You can also read about the MA plot in DESeq2 by reading Section 1.5.1 MA-plot, in the DESeq2 vignette:

vignette("DESeq2")

Or the help page for DESeq2's plotMA function:

?plotMA

Or the description of DESeq2's MA plot in our RNA-seq workflow:

http://www.bioconductor.org/help/workflows/rnaseqGene/#plotting-results