Generally, it doesn't really matter unless you're publishing a custom, novel method where the code is really an integral part of the publication. In such a case, you'd want to keep it private until publication. Otherwise, keeping your random DESeq2 scripts in a public repo isn't going to hurt you or draw much attention, though nobody will care if you keep it private until submission either.

Just be sure not to include any PHI or sample information in them.

I would strongly recommend publishing them before the paper. That way, you can get your code peer reviewed before you submit your article. That may improve your code. You may have to arrange reviewers yourself for this (maybe an idea for a new BioStars section?).