Ab initio solution of metalloproteins, warp_solve.sh

If the resolution of your data is lower than 1.5 Å, FORGET IT! And by 1.5 Å we mean really good 1.5 Å data! Our test sets were truncated to this resolution which is obviously not the same as if the real data actually stopped there. In general, data to 1.2 Å is typically sufficient enough. If, however, you have a few heavy atoms and good data to a resolution of about 1.2 to 1.5 Å, it won't hurt trying.
The script warp_solve.sh will take care of everything. The only requirement is that you provide the heavy atom positions in Brookhaven format, these should be placed in a file called heavy.warp.

A possible heavy.warp might look like this:

ATOM	51  SG  IUM X	6	0.217	0.120	8.404	1.00	5.21	8RXN  92
ATOM	71  SG  IUM X	9	3.812	0.666	8.019	1.00	6.22	8RXN 112
ATOM	306  SG  IUM X	39	2.881	-2.301	9.553	1.00	5.36	8RXN 347
ATOM	326  SG  IUM X	42	2.226	0.925	11.274	1.00	6.36	8RXN 367
ATOM	399 FE   IUM X	55	2.222	-0.095	9.294	1.00	5.14	8RXN 440

Notice that this follows standard PDB file conventions except that the CRYST and SCALE cards are not needed here (and should not be in there).
Run warp_solve.sh and hope for the best... In case that ab initio does not work, do not forget the option to perform a SAD (Single-wavelength Anomalous Dispersion) experiment. Even when the data are collected far from the optimum wavelength for maximising anomalous scattering, most metals have an appreciable signal. Try to get some (even weak) phase information with another program and go a few pages back to read about mode warpNtrace.