Normalised counts of control samples in gene expression matrix

Question

[Pranavathiyani G]

I have 4 control and 4 treated samples of RNA seq data, for generating gene expression network should i consider adding the normalised counts of control samples in the gene expression matrix?

[cpad0112]

You are supposed to calculate DEGs or Statistically significant genes between two groups (control vs treated) using methods of your choice (in general people use scientific community accepted packages such as ballgown, DESeq2, EdgeR etc). For this analysis, you would need raw counts than normalized counts. Then one can plot expression networks. Just with one group (in tihs case control samples), it doesn't make sense to me, to arrive at expression networks.

Answer 1

[Devon Ryan]

Yes, you would tend to include all of the group when making a coexpression network. You need larger sample numbers to make a reliable network anyway, so you can only achieve that by including all of your samples.