Ideas for DDP (cont’d)
A low level scoring matrix ijR is defined for each pair (ai,bj)
ijRkl is a score showing how well the residue ak fits to bl under the constraint that ai is aligned to bj.
Low level DP is performed using ijR (in principle for each pair (i,j))
The contribution from ijR is all ijRkl such that (ak,bl) lies on the optimal (low level) path