I'm a postdoc that works mostly with biochemist-ey types, and I'd highly recommend adding a math package to whats available to your students. With something like mathematica, you can do:
- curve fitting/minimization
- general math (probability distributions, convolutions, Fourier transforms, etc)
- peak integration
- statistical analysis
- image analysis
- check formulas and their characteristics (being able to verify derivatives, for example, is very important for fitting models)
what I also like about the math packages is the ability to synthesize "test" data to illustrate what can't be done simply in lab (or not at all, depending). And I think its also a great way to start learning a bit of programming/scripting without requiring too much CS (for those not interested in CS), but at the same time getting enough exposure to it so that they won't be completely lost when they see a conditional loop. And, I can personally tell you that science types use them quite widely.
I'm a little surprised you seem more concerned about the OS the programs run on. As long as the students can run the stuff you've listed along with some sort of math package to learn about handling data, just go with what the IT guys are most comfortable with.
But to answer your question: most science labs run whatever they want, but some hardware and/or proprietary analysis software for some equipment can dictate the OS.