Where to start.... The simplest answer is "some other"
If you are not satisfied with this answer, I would first try reading the paper though it might be hard to understand so it could be easier to start with a review.
You can think of what DESeq does as kind of like a modified t-test on counts but the counts need to be normalized before you can use them. Istvan's answer here provides a great starting point: https://www.biostars.org/p/99689/ as well as this post
Assuming you have a good understanding of NGS count data, you should also learn how linear modeling work. The user guide here is great for walking a beginner through how differential expression analysis works. edgeR is a tool similar to DESeq though there are some differences.