Geometric and Semantic Analysis

From the PDB and topological analysis, we construct a molecular graph with untyped bonds and recognized chemical rings. For correct chemical recognition, we need to determine hybridization states, from which we then try to derive correct bond types. 

The first assignment step was to detect planar rings that obey the Hiickel rule of 4n+2 p-electrons and flag them as aromatic, i.e. force all the atoms into an sp2 hybridization state. For planarity recognition, three arbitrarily chosen adjacent ring atoms form a base plane. Subsequently, for each atom of the ring, the distance to this base plane is computed, d being zero for the ideal case. In vector notation, the plane equation can be described as follows  

As rings are rarely perfectly planar, a threshold for the in-plane-distance had to be determined Comparing distorted pyrroles in protoporphyrin rings with sp3 atoms in pyrrolidines reveals that a threshold of 0.65 A is adequate for the determinant described above. 

---