I have RNAseq results from NovaSeq and after doing some differential gene expression with DESeq2, I will like to explore the Elastic Net model to select features that differentiate my two conditions.

However, I haven't been able to find a nice workflow or guidance about how to run this analysis. So, I have a couple of specific questions.

• Should I employ VST corrected gene expression as an input for the elastic net model
• Any suggestions about how many hyperparameters (lambda) to evaluate
• My dataset is small and this is exploratory, so I plan to do LOOCV (Leave-one-out cross-validation), any thoughts about it
• Anyone knows of a tutorial, repository or such with start to end Elastic Net for RNA-seq data

Thanks

We recommend VST for downstream ML methods. Don't have any particular advice for ML downstream, but there are some packages on Bioc:

https://bioconductor.org/packages/release/BiocViews.html#___Classification

E.g.

• https://bioconductor.org/packages/release/bioc/html/MLInterfaces.html
• https://bioconductor.org/packages/release/bioc/html/MLSeq.html