Benchmarking different DESeq2 design models

Question

[jdhindsa1999]

Hello,

I have RNA time-course data and am looking to benchmark two different DESeq2 models I want to use. One design models time as a categorical variable, while the other design models time as a continuous variable. I am wondering if anyone knows of any R packages or methods to compute values such as AUROC to compare these two models performance in detecting DEGs.

Thank you

[jv]

Do you have a set of validated DEGs (from another source of data) that can serve as the ground truth for your DESeq2 models? Is this in-silico RNA-seq data or real experimental data?

Answer 1

[jv]

To generate AUROC statistics requires both a set of defined true positives and true negatives to serve as the "ground truth" to compare model performance against. This is typically achieved via cross validations when developing models.

For an example from a related post see https://www.biostars.org/p/271616/#271616