Normalization is essential for RNAseq analysis. However, one needs to understand the underlining assumptions for each methods. Most methods assume there is no global changes between conditions (e.g. TMM normalization). However, this may not be true when global effect occurs. For example, if you delete a gene that controls transcription, you expect to see global gene expression reduction. In that case, other normalization methods need to be considered. (e.g. spike-in controls). The same principle applies to other high-throughput sequencing data such as ChIPseq. To estimate the library size, simply taking the total number of (mapped or unmapped) reads is, in our experience, not a good idea. Sometimes, a few very strongly expressed genes are differentially expressed, and as they make up a good part of the total counts, they skew this number. After you divide by total counts, these few strongly expressed genes become equal, and the whole rest looks differentially expressed.