What is the cutoff distance in this algorithm?

Question

In the following article:

• a fast and efficient Cα-based method for accurately assigning protein secondary structure elements

What are the minimum and maximum cutoff distances in this algorithm?

Also, what does it mean by:

I.e. where does (2x43)x3 come from?

Answer 1

The 43 features are listed in the supplementary materials on page 4. For example,

Feature Type Atoms Involved 
1 Distance i, i-5 
2 Distance i, i-4 
3 Distance i, i-3 
4 Distance i, i-2  
...
11 Angle i-1, i, i+1 
12 Dihedral i, i+1, i+2, i+3 

Are some of them; so Feature 11 is the angle between atom(i-1), atom(i), and atom(i+1). Since it says these were calculated for both the C-alphas and pseudo-centers, this gives 2 * 43. Then this is multiplied by three as mentioned in your quoted paragraph.

These features (measurements, almost) are numbers, not booleans - so I do not think that they use distance cutoffs as such, although the random forest method is presumably learning various cutoffs and encodes those in the structure of the forest.