A word of caution

QS is not the program to use as a first attempt at a molecular replacement problem. It is rather ``yet another program to try'' when you are desperate enough and all else had failed. There are several reasons for that :

1. QS is slow. Although for a typical problem --and on a modern workstation-- the program can examine several hundred configurations per second of CPU time, the dimensionality of the configurational space guarantees that you will probably need tens of million moves if you want to have any chance of finding the global minimum of the target function (which, after all, may not coincide with the true crystal structure).
2. QS, being based on a Monte Carlo minimisation, is by nature stochastic, and so there is no guarantee that even if a pronounced global minimum exists, it will be found during your lifetime. The good news (if you are an optimist) is that if a pronounced global minimum does exist, there is a slim chance that your next minimisation will fall inside it three hundred moves after you submitted your batch job.
3. The CPU time required per move is proportional to the product {{ Number of unique reflections } x { Number of symmetry operators }}. The good news is that the CPU time per step is independent of the number of atoms in the search model and of the number of molecules in the asymmetric unit. The bad news, is that if you insist on using all your reflections to 2Å for the minimisation, you better start the batch job just before your summer vacations (and I bet you won't be very popular upon your return). ⁶
4. QS is very demanding on hardware resources, especially physical memory. If your workstation has less than 64 MBytes of RAM, you won't be able to use QS with molecules bigger than (approximately) the size of a lysozyme molecule (and not at a resolution higher than about 4Å). You also need a fast CPU with a floating point co-processor if you want your simulations to finish this week (needless to say that you need a batch queue without time limitations).

Footnotes

... return).⁶

In other words, the CPU time per step is proportional to the volume of the unit cell (if you double the length of all cell edges, the CPU time per step will go up by a factor of eight).