I’m a biochemist, so have only grown a few species of bugs. I always call E. coli the workhorse of the lab, since they grow fairly quickly.
The software modelling comes in handy when you are dealing with proteins that are too unstable or cannot be expressed in high enough quantity to crystallize. Some proteins do not crystallize. Back in grad school, during one of our lessons on protein structure analysis, the professor told us how he would spend weeks entering the characteristics of a protein into a code, which he would then take to UCLA and, for $16,000/hr computation time (20 years ago), he could feed the code into a CRAY and get some modelling results. He had to be very careful about the programming; at that price, he did not want a buggy or incorrect code. For all that money, the output was a 3D line diagram of a protein vibrating between its possible conformations, lasting a few seconds.
When I still lived in San Diego, one of my nearby neighbors was in the process of creating a company to do the protein modeling you described. That was in the very early 90's. It appears he succeeded and left the immediate area to tend to his new company. It certainly makes more sense to model that way.
I agree with E. coli being the workhorse in the lab. The EcoR1 mutant (no restriction enzymes) was a big step forward. That was barely available when I graduated in 1976.