Ideas For Best Representation Of Differentially Expressed Genes

Question

[Dataminer]

Hi!

I have three RNA-seq samples and by pairwise comparison I have computed differentially expressed genes between them.

I have two questions:

(i) Is there a way to compute the differentially expressed genes in three samples simultaneously rather doing pairwise comparison and getting them?

(ii) what is the best way to represent these differentially expressed genes, for differentially expressed genes between two samples can be represented by a scatter plot but is there any better way also?

Thank you

Answer 1

[David]

For extracting DE gene across 3 conditions you can use an ANOVA procedure to differentiate between different models or limma and voom (tutorial). A third approach would rely on clustering expression profile based on a parametric or non-parametric correlation metric of your choice. For example applying a Self-Organizing Map or a k-Means. However those method will not provide significance level for the genes.

For representation purpose you can always have the expression on the Y-axis and the conditions onthe X-axis. If you want to go further hive plots looks cool (http://mkweb.bcgsc.ca).