Biologists don't necessarily need to be the one writing the high-end software tools, in the same way (as you note) that they don't build their own microscopes.

But re-formatting, filtering and running basic QC on complex data is something a lot of biologists have to do every day; and the types of data, their quality and their format are constantly evolving. Biologists who have to run down the corridor to their local (probably overworked and underpaid) programmer every time their data format changes, then explain the problem in terms the programmer can understand, then wait for the programmer to write something, and then check the script's output for errors, will be at a huge disadvantage compared to those who can quickly whip up and debug a parsing script themselves on the fly.