Limma for RNA-Seq between two pre-defined clusters.

Question

[simran.ven]

I am a complete beginner at 'R', so bear with me. Currently, I have an RNA seq data set that identifies two clusters (High exp samples and Low exp samples) based on a gene set. I now want to explore the differential expression of the remaining 20k genes between these two clusters. My advisor has told me to use limma, and I found a pretty useful link : RNA-Seq analysis is easy as 1-2-3. My approach was to conduct the limma analysis separately for the two clusters. Half way through my attempt I began to question if that was right, as the article kept mentioning DGE compares between conditions..

My question is: How do I prepare my RNA-Seq data. Currently, it is framed as Rows: Samples Columns: Genes How do I factor in the two clusters in this data frame, to use for DGE? If there is a complete dumbed down version, please do tell. Any help is appreciated.

Answer 1

[Kevin Blighe]

The components of any expression study are usually:

• expression matrix / matrix of raw counts (genes as rows; samples as columns)
• metadata (samples as rows; phenotypes as columns)

The order of the metadata rows should match that of the columns of the expression matrix.

You can compare anything that you want in a differential expression analysis. Condition is a term that just happens to be used a lot in tutorials.

When I was a beginner in Bioinformatics, I went through those types of tutorials on my own, making my own comments in my code. Once completed, it then served as a 'prototype' that I could then apply to my own data. I suggest that you take this approach. Take the time to go through the tutorial, and using the data that they use in the tutorial itself.

Kevin

[simran.ven]

Thank you for being kind enough to explain the crux of this. Yeah, I understand that it is necessary to take the time to go through tutorials, but it's frustratingly slow. I guess I will have to go through it any which ways. Thank you again!