Required crystal number or size calculator

nxtals = <IDL> / 20 * fNH * MW * VM2 / exp( -0.5 * B/reso2) / xtalsize3 / ( reso3 - 1.53 )

Enter values:
experiment goal =
number of sites = in asymmetric unit
fpp = electrons Bijvoet ratio = %
molecular weight = kDa in asymmetric unit
resolution = Ang signal to noise = at this resolution
reso on snapshot = Ang → Wilson B = Ang2
background level = ADU/pixel multiplicity =
spot size = pixels
detector type =
detector gain = ADU/photon
blank image level = ADU blank image rmsd = rms ADU
solvent content = %
xtal sizebeam = microns
xtal sizevert = microns beam sizevert = microns
xtal sizespindle = microns beam sizespindle = microns
nxtals = xtals you will need to merge ← <IDL> photons/hkl

Note that you want nxtals to be smaller than 1 for the experiment to work. Preferably much smaller. If nxtals is around 1 or greater, then you will probably need to merge data from a number of crystals. The text box will change from red to yellow to green to indicate this.

Changing nxtals above will update the crystal size appropriately. Note that the crystal size is entered in three directions: along the spindle rotation axis (sizespindle), and the directions swept out by the beam as the crystal rotates (sizebeam and sizevert). sizespindle may not be larger than the beam because parts of the crystal outside of the beam do not contribute to the diffraction pattern. If your crystal is actually longer than this, then you can merge data from several parts of it, which is indicated by nxtals > 1.

Changing the field for observed "reso on a snapshot" will update the Wilson B factor to a reasonable expected value from PDB-derived statistics. This B value will be appropriate if you enter the resolution of the faintest visible spots on a fairly long-exposure diffraction test image from your crystal.

The "background level" is the pixel intensity you see near spots at your resolution of interest, and the units of "ADU" are the pixel intensity units in your diffraction image display program. The change in "ADU" generated by an average of one photon/pixel is called the detector "gain", and this is the same value you input into MOSFLM. The "gain" is not always 1 (in fact, it is usually more than one), and depends a lot on the detector type and on the photon energy. Typical values for popular detectors may be selected with the pull-down menu, and arbitrary detectors may be programmed in with the "Custom..." option. The values for "blank image level" and "blank image rmsd" can easily be obtained from an image taken from the detector with no x-rays.

<IDL> is the damage-limited merged spot intensity (Holton and Frankel, 2010). The required value for the experiment to work depends on the noise level, and the needs of the experiment. For high-angle data (weak spots) you generally need a signal to noise rato (snr) of about 2, but for MAD/SAD data, you need to measure a small difference between bright spots. The noise level, snr and <IDL> will be updated upon changing parameters that affect them, such as spot size, Bijvoet ratio, etc. Entering zero heavy-atom sites indicates you intend to do high-angle data colleciton.

fNH is the Nave-Hill capture fraction ( Nave and Hill, 2005, Holton and Frankel, 2010), which becomes less than 1 as the size of the beam and crystal in the "spindle" direction becomes small when compared with the photoelectron track length (about 3 microns for 12 keV photons).

Changing the number of sites or the anomalous electrons will calculate a new Bijvoet ratio, and this, in turn will update the signal-to-noise and number of photons required for the experiment to succeed.


For more information see:
Holton, J. M. (2009). "A beginner's guide to radiation damage." J. Synchrotron Radiation 16: 133-142.

Holton, J. M. and Frankel K. A. (2010). "The minimum crystal size needed for a complete diffraction data set" Acta Cryst. D66: 393-408.