I have read some papers where the authors perform simulations at constant pressure and constant volumen or sometimes constant volume constant temperature. My question is when is it better to use one or the other? For which systems it would be ok to perform constant p-t and for which would be better constant v-t?
The choice doesn't depend on the system but on the properties you want to measure (which means that it depends on the computational experiments you need to perform).
In NPT: Pressure is constant. Volume (and density) is changing.
In NVT: Pressure is changing. Volume (and density) is constant.
So, for example, if you want to plot a property A of your system vs pressure at constant density (or volume) then you use NVT. If you want to plot property A vs volume (or density) at constant pressure, then you use NPT.
This question is not really answerable as it stands, because "it depends".. a good rule of thumb is that you should consider what the real system you're modeling is doing.
In a cell, this would probably be constant pressure, constant temperature (an NPT-ensemble).
However, when you construct molecular dynamics setups you need to minimize and equilibrate structures all the time, and then you might need other regulators. For example if you have managed to create a vacuum bubble somewhere in your structure and you hope to let water fill it in and smooth out your fiddling, a constant pressure regulator could go berserk and then you'd probably start with a constant volume constant temperature (NVT) regulation.
If you're doing this at a university, ask around, I'm sure you'll hear as many equilibration protocols as the people you ask :)