The problem is that some interactions (bonds, angles) are symmetric, others are sometimes (dihedrals) and others are not. And note that for angles the order of nodes matters anyway.
Although, after some discussion, this might actually be solved by using ISMAGS. To restore the transitivity of node equality however, the 'order' operation has to go. And it can, since it's designed to solve the same problem.
If interaction asymmetry becomes an issue the link is underdefined anyway.
The problem is that some interactions (bonds, angles) are symmetric, others are sometimes (dihedrals) and others are not. And note that for angles the order of nodes matters anyway.
Although, after some discussion, this might actually be solved by using ISMAGS. To restore the transitivity of node equality however, the 'order' operation has to go. And it can, since it's designed to solve the same problem. If interaction asymmetry becomes an issue the link is underdefined anyway.