Not that different from 3.0: the main change is to allow control over various part score importances like hiding.

Algorithm outline

For each amino acid
Set the Clash Importance to something low
Select everything in the immediate neighbourhood of that residue
Mutate
Set the CI to 1 and select the whole protein
Wiggle
Record any improvement
end

Dialog settings

Min residue, Max residue

The range to be mutated : by default the whole protein with locked ends trimmed off.

Delta residue

Processing each residue in turn is inefficient: the selection in each case will be largely duplicated. This setting indicates the difference between successive residues. For example, if the min and max residues above are 1 and 100 respectively and the delta value is 5, then residues will be processed in the order 1,5,10,15,etc.....2,6,11 and so on.

Clash importance

The value during the initial mutate. Something that reduces the energy score by about 200 seems about right (the median reduction is printed out at the end of the script)

Modify count

Determines how many amino acids will be mutated at each step. They will be the closest (as measured by the distance between alpha carbons) to the residue being processed. Fewer means faster processing and will enable finer control over the mutating process: for example it may be desirable to minimize the chance of a mutation disrupting any existing Hydrogen Bond Networks.

Packing importance
Hiding importance
Pairwise importance
Backbone Hbond importance
Sidechain Hbond importance

Sliders added to control these values (in Other Options). If you change them make sure Behavior->Recipe Score Modding is checked. Have you to get a handle on how useful changing these will be.