Re: [modeller_usage] about random seeds

On 3/6/11 9:03 AM, albert wrote: > I am using the random seeds when I building model these days. and I > found that in the following line: > > environ(rand_seed=-8123, restyp_lib_file='$(LIB)/restyp.lib', copy=None) > > I try three jobs with rand_sees=-500, -1000, -1500 > > and I found that the -1500 get the lowest DOPE score and -500 get the > highest DOPE score for the best solution. > Does it means that if I set rand_sees=-2000 would get a better one?

No. There is no correlation between random seed and model quality - it is random.

Ben Webb, Modeller Caretaker