Wednesday, April 24, 2013

Visualizing a lot of FMO PIEDA numbers. Part II

In the last post on the subject of FMO PIEDA numbers, there were some issues with the fragmentation that did not match between two different snapshots in the same reaction trajectory.

I fixed that fragmentation problem, ran the numbers again and it looks like some neat stuff can be extracted.

Remember that the structures are generated using PM6 with MOZYME enabled and that energy-refinement of the barrier is done at the FMO2/6-31G(d)/PCM level of theory. The CON structures have some part of the structure constrained (far away from the active site) and the UCO structures are allowed to fully relax. The problem still looks like this



To tease you with some data, here is a table from a short write-up I did. Column two and three represents the difference in energy (by the method in the first column) between the TS and the reactant. The last column is the energy difference between column two and three. The first row is thus the difference in sum of one-body energies, second row is difference in the sum of fragment interaction energies (FIEs) and the last row is the difference in total FMO2 energy.

dE(1,5)_con dE(1,5)_uco ddE(uco-con)
FMO1-MP2/6-31G(d)/PCM 25.4 29.4 4.0
$\sum_{IJ} \Delta E^{MP2}_{IJ}$ -3.7 18.6 22.4
FMO2-MP2/6-31G(d)/PCM 25.9 55.0 29.1

The one-body case is a solved case because inspection reveals that the internal energies of the substrate contributes +3.5 kcal/mol, ASP119 contributes +5.3 kcal/mol and ASN35 contributes -4.3 kcal/mol which when summed up is roughly the 4 kcal/mol we need. The rest is just internal re-arrangement that eventually cancels out.

The problem with the barrier height, however, is that at the two-body level (second row), the CON structures provide almost 4 kcal/mol worth of stabilization whereas the UCO structures are destabilized by 18.6 kcal/mol. It would be natural to investigate what happens if we look at the FIEs between the substrate and the entire enzyme. This looks like the following


what is curious is that we see distinct peaks - the largest (positive) peak can be attributed to an interaction with ARG112, but if we sum all IEFs up they amount to +0.5 kcal/mol which is very far from the +18.6 we are trying to account for. The conclusion is (currently) that because the protein is allowed to fully relax, many small contributions amount to the 18.6.

credits to this sites table-generator because I lost all my HTML skills whilst fighting FORTRAN