Entering edit mode
3.9 years ago
Hi,
I'm try to do weighted gene co-expression network with WGCNA. I have 13 mouse samples, one condition, I know it's not ideal to perform WGCNA, as they suggest more than 15 samples.
I tried on my samples, there's an issue with the scale free topology index, my scale is from -0.3 to 0.15. That's very odd, I don't know because I used only 13 samples or because something's wrong with my pre-processing part.
Thank you in advance for your help!
Yi
Hi Gabriella,
We can't guess what is wrong with your data without knowing how these have been pre-processed. Also, testing 13 samples representing only one condition, without any trait data, doesn't really help you to understand the biological relevance of your modules
Hi Andres,
I normalized bulk RNA-seq data with DESeq2 and also did the variance-stabilizing transformation which the tutorial suggested. I have no bioinformatics background, so I don't have very good judgment about choosing the right method.
Because I'd like to do some module analysis based on the publication https://www.embopress.org/doi/epdf/10.15252/embj.2019101828, then I realized that they used TF-IDF transformation, I can't find any tutorial for it. So I decided to use WGCNA which is also for module analysis.
If you have any thoughts or ideas, please let me know. I really appreciate it!
Thank you in advance! Yi
DESeq2 transformation works fine for bulk RNA-seq data. In regard to the work you just linked, they use a normalization method specific for single cell RNA-seq data, and a transformation method (TF-IDF) that suppress the signal of widely expressed genes. I do not think these methods can be applied to your data set.
As I said, testing only13 samples representing one condition and without any trait data it is not optimal for WGCNA.
Thanks. I'll take other approaches to analyze the data then. I much appreciate your help!