You'll need a dose response curve, from which you can calculated IC50's using a variety of free (R, NCGC CurveFit, curve-fit) or proprietrary tools (Prism etc)

One approach to calculate IC50 values, is to do so by comparison. Similar molecules can be expected to have similar IC50 values. For this, QSAR studies can be used, and have extensively been used (filter through merely related papers). These kind of studies can also include information from the protein structures, e.g. with proteochemometrics approaches.

In my view, IC 50 value should be obtained experimentally in stead of deriving it from QSAR study. Even though similar molecules can be expected to have similar IC50 values, I have done extensive study on this issue and have inferred that, different experimental procedure result in different IC 50 values for the same chemical compound. And the difference are not trivial.