Structural determination of large molecules through the reassembly of optimized fragments

The accurate determination of the optimized structures of large molecules is, computationally quite expensive. As an alternate to the conventional approaches to structural optimization, we have explored the accuracy and speed-up obtained when variants of the fragmentation optimization and recombination method (FORM) are used. Specifically, the method was applied to eight prototypical molecules -n-decane, hexa-alanine, a long conjugate hydrocarbon molecule, a large polar conjugated molecule, large (5,5) armchair single-walled carbon nanotubes, a salen-aluminum complex and a multiply H-bonded system (two conformers of vancomycin aglycon with Di-N-acetyl-l-Lys-d-Ala-d-Ala - without optimizing the structure of the whole molecules. We find that FORM can predict the structure of these molecules with an acceptable accuracy, all at a computational cost that is 2-11 times less than conventional quantum mechanical methods at the Hartree-Fock (HF), density functional theory (DFT) and MP2 level of accuracy. FORM may therefore represent a viable approach for the fast structural predictions of large molecules.