Hierarchical Clustering And Heatmap Analysis For Rna-Seq Data

Question

[KJ Lim]

Dear BioStar community,

Good day.

I have analyzed my RNA-Seq data with edgeR (R Bioconductor). The differential expression (DE) gene lists of my RNA-Seq data were obtained.

I would like to perform hierarchical clustering as well as heatmap analysis for the DE lists I obtained. I have been searching around for the right R packages or tools for these purposes as the RNA-Seq data is not same as microarray data.

Could the community kindly please light me where and how I should start?

Thank you very much. Have a nice day.

Answer 1

[Obi Griffith]

I don't think you necessarily need any special packages for doing clustering/heatmap analysis of RNAseq data versus microarray analysis. I've used heatmap.2 {gplots} for both. It has hierarchical clustering built in and you can easily specify different distance metrics and/or linkage criteria. You can start with your expression matrix, extract the subset for differentially expressed genes, and feed that into heatmap.2. If you search biostar for "heatmap" you will find all kinds of suggestions and tips on how to proceed as well as example code.

[KJ Lim]

Thanks for the input :)

[cmdcolin]

Is there any particular wisdom to only selecting differentially expressed genes? It makes sense but I was wondering if this is only done because it is RNA-seq instead of a microarray, or maybe it was done with microarrays as well?

Answer 2

[Nicolas Rosewick]

Check the DESeq package vignette (bioconductor) . There is a whole chapter on clustering RNA-seq data. http://bioconductor.org/packages/release/bioc/html/DESeq.html

[KJ Lim]

Dear Nicobxl, thanks for your suggestion. My RNA-Seq data already analyzed with edgeR, could it possible integrated in the DESeq package?

[Nicolas Rosewick]

DESeq used the same input as edgeR : a read count matrix. So yes it's very easy ;)

Answer 3

[Damian Kao]

I wrote a post on this few weeks ago on how to do hierarchical clustering in scipy and displaying the results in a browser with javascript: http://blog.nextgenetics.net/?e=44

You can also adapt it to display the heatmap in R with ggplot2's tile plotting functions.

[brentp]

pretty cool. would be nice if you could wrap that all in a function that would output the .js or fill and html template.