With the exception of modifications for full-structure data sets having a resolution lower than 1.1Å (section 6 below) the following recommended parameter settings (as of this release) are built into the program as default values. A current list of parameter recommendations is maintained on the SnB web site. Several parameter values are dependent on "Nu", the number of unique non-H atoms in the asymmetric unit. In the case of solvated molecules, it is generally not critical whether the solvent atoms are counted when determining Nu. In the case of SIR or SAS substructures, Nu is the number of heavy atoms or anomalous scatterers.

1. Normalization Parameters
   • Use Bayesian estimates for weak reflections? Yes
     • Do NOT do this if weak reflections have been previously zeroed or replaced by threshold values.
   • Use locally normalized |E | values?
     • No - if the maximum resolution is better than 3.5Å
     • Yes - if the maximum resolution is worse than 3.5Å
   • Perform local scaling? No
   • |F | / sig(|F |) cutoff for local scaling: 3.0
   • Difference |E | limits:
     • Tmax: 6.0
     • Xmin: 3.0
     • Ymin: 1.0
     • Zmax: 0.0
     • Zmin: 3.0 (applied in main phasing program)

2. Phase Refinement Parameters
   • Phases: 10*Nu
     • |E |/Sig(|E |) cutoff (Zmin): 3.0
     • Max |E |: 5.0 (for small centrosymmetric structures, it would be reasonable to allow for larger |E | values)
   • Triplet structure invariants: 100*Nu

   • Phase refinement method: Parameter Shift
     • Space Group: Noncentrosymmetric (except P1)
       • Phase shift: 90°
       • Maximum number of attempted shifts per phase: 2
       • Complete passes through the set of reflections: 3
     • Space Group: P1
       • Phase shift:
         • 90° if Nu < 100
         • 180° if Nu > 3892
         • 22.5*ln(Nu) - 6, otherwise
       • Maximum number of attempted shifts per phase: 2 if Nu < 100, 1 otherwise
       • Complete passes through the set of reflections: 1
     • Space Group: Centrosymmetric
       • Phase shift: 180°
       • Maximum number of attempted shifts per phase: 1
       • Complete passes through the set of reflections: 1

3. Dual-Space Fourier Parameters
   • Peaks Recycled:
     • Nu if Nu < 100; else
     • 0.4*Nu if atoms heavier than oxygen are present or
     • 0.8*Nu otherwise
   • Fourier grid size: (Resolution of data in Angstroms)/3
   • Minimum interpeak distance: 1.0Å
   • Minimum distance between symmetry-related peaks (used to define the excluded volume about special positions): 3.0Å
   • Peaks permitted at special positions: 0
   • Perform extra cycles with more peaks? No (but see Fourier Refinement screen)

4. Twice Baking (E-Fourier Recycling) Parameters
   • Best trial only
   • Number of cycles: 0.1*Nu
   • Number of peaks: Nu if Nu < 100, 1.1*Nu otherwise
   • Minimum |E | : 0.75

5. Trials & Cycles
   • Trial structures processed: 1000
   • Source of starting phases: Randomly positioned atoms
   • Seed for random number generator: Use a 5-digit prime number (default 11909)
   • Cycles:
     • Nu/2 if Nu < 100 or Nu < 400 and atoms heavier than oxygen are present
     • Nu otherwise

6. Using SnB at 1.1-1.2Å
   Note: These changes are not made automatically
   • Increase the number of invariants (200*Nu - 500*Nu) and/or
   • Perform more cycles (2*Nu)

7. SnB for Substructures (Nu = number of heavy atoms or anomalous scatterers)
   • Phases: 30*Nu
   • Invariants: 300*Nu
   • Peaks: Nu
   • Cycles: 2*Nu (min. 20)
   • Minimum interpeak distance: 3.0Å
   • No twice baking