On Fri, Mar 1, 2013 at 12:43 PM, Michael Thompson
I am trying to simply copy a set of FreeRFlags from a higher resolution dataset to a lower resolution dataset. The low-res dataset contains 35264 reflections in the original input mtz. The high-res set contains 43731 reflections. When I use the GUI to write a new mtz using the low-res amplitudes and the high-res FreeRFlags, the result is an mtz with 46323 reflections. I am confused about why the output has so many reflections, even more than the high res dataset? Also, I noticed that while both input mtz files have a low-res cutoff of 20A, the output mtz has a low-res cutoff around 75A. Can anyone please explain what is going on here? I would like my output file to have the same reflections as the input, and no more.
I need to double-check my code, but I'm pretty sure it generates flags for all possible reflections up to the specified resolution limit when you choose the extend option. (The theoretical low-resolution limit is determined by the unit cell parameters.) There was at least one specific (and somewhat complicated) reason for this, but the most important is simply that there is zero advantage to doing it the other way. None of the programs in Phenix care whether you have more flags than data - they will simply filter out the surplus flags.
I have noticed that the CCP4 program FreeRFlag seems to behave in a similar way, so my guess is that what is happening is a common procedure, but I'm at a loss for an explanation of why this is the case. Also, if there's a better way to implement what I am trying to do I'd be grateful to hear it (perhaps generate entirely new flags for both datasets using the same random seed?).
I guess I can add an option to strip out the extra flags, but it's unclear to me why this would actually be necessary. If a program is crashing because the data and flags arrays aren't the same size, that's a bug. -Nat