The idea of the method is to find all the
compatible (superimposable) secondary structure
elements, and then to form the large ensembles
of mutually compatible elements.
Search for the compatible pairs of the SSEs
For each pair {Si,Sj} of the SSEs from protein A, we calculate:
- the angle Gij between them,
- the shortest distance between their axes,
- the corresponding closest points on the axes (Cij on the axis
of the Si and Cji on the axis of the Sj),
- the minimum (Dmin)
and maximum (Dmax) distances from each SSE to their medium line.
The medium line is defined by the middle point of the shortest
segment between axes of the SSEs and the vector (Si/|Si|+dij*Sj/|Sj|),
dij = +/- 1. We chose the sign of dij to
place projections of the vectors Si and Sj to the medium line on one
side from the projection of the closest points. In the case when
vector Si goes through the closest point, we devide it into two parts,
separated by the closest point Cij. Totally we have 5 parameters to
determine the compatibility of two pairs of SSEs: angle Gij, and two
distances for each of the SSEs (Dimin, Dimax, Djmin, Djmax).
We consider pairs of the SSEs
{SAi, SAj} from protein A and {SBk, SBl} from protein B as
compatible, if :
- |GAij - GBkl| < G0 (usually G0 = 50 degrees)
- Dimin - Dkmax < e ( e = 1.5Å )
- Dkmin - Dimax < e
- Djmin - Dlmax < e
- Dlmin - Djmax < e
These distance and angle constrains filter out differently arranged
pairs of SSEs, leaving only superimposable SSEs. In contrast with
the distance constrains, based on the distance between centers of
the SSEs, our filtration procedure can detect difficult cases, when
only relatively small segments of the SSEs are compatible, or when
one of the SSEs is significantly larger than another.