Why we need computational methods for predicting DHS activity?

Question

[Sama]

In this paper (Basset), the authors try to predict DHS activity. My question is that why they predict DHS? Is experimentally extracting DHS activity expensive? They just explained that it is useful to have DHS activity but they did not mentioned what is the point of having a computational approach for that.

[Jean-Karim Heriche]

I can imagine situations where you don't get enough material to do additional experiments after sequencing or where you need this information when you don't have access to the samples (e.g. data from repositories).