The constraints just mean limitations of the area over which an event is unpredictable ("perfectly random"). Let me give you an example. Aflotoxins cause DNA mutations by binding to parts of guanine residues resulting in a G to T transversion. The toxin is very biased in this affinity, because it only binds to G and not to other nucleotides. At the same time, the toxin has no affinity for one available G over another G. It is constrained by chemistry, not by a directed or goal-oriented process, and, as such, it is still unpredictable within those constraints.
And this proves what?
I guess there is no constraint to the randomness of information if there is no constraint to usefulness of information.
But that is what you meant when you stated in regard to DNA:
The constraints just mean limitations of the area over which an event is unpredictable ("perfectly random").