I recently watched a presentation on RNA-seq that covered some of the choices one can make along the way, and I didn't fully understand one of the choices in particular. Near the beginning of the process, you can choose a fragmentation method (e.g. RNA: nebulization, hydrolysis; cDNA: sonication, Dnase I treatment). Each fragmentation method is biased in its own way, and the goals of the study may influence the choice of method.
I didn't fully understand the reasons for the characteristic biases (how do they happen?), and when to choose one method versus the other (which methods work best for which goals?). Put another way, what are the mechanistic explanations for the characteristic biases that result from each method, and what is the significance as far as downstream processing and associated challenges?
The heart of this question is about fully understanding the particular choice of when and how to fragment. The reason I have not accepted an answer, yet, is that the answers provided so far address the mechanism part of the question, but not the ultimate reasons for choosing one versus the other.