Visualizing a lot of FMO PIEDA numbers. A Preliminary look.

I've assisted in calculating some energy refinements using FMO2-MP2/PCM/6-31G(d) on top of a colleagues proposed trajectory calculated using PM6//MOZYME for part of the reaction step of Bacillus circulans xylanase. Two versions of this path was produced. One with constraints on some part of the structure (CON) and one without constraints (UCO). They are shown here

This FMO2 barrier is quite unrealistic, especially the unconstrained one so what is cause this large reaction barrier?

To investigate this, we are trying to run some PIEDA calculations to figure out which pairs are interacting strongly (and perhaps differently). PIEDA gives you an energy decomposition analysis of the individual pairs in an FMO calculation. So we get electrostatic contributions, exchange-repulsion, charge-transfer (and what is left), dispersion and solvation energy too.

The BCX-system we are looking at currently has 302 fragments (a total of 3172 atoms) which is actually the whole protein and its substrate. That means you get 45451 unique pairs and each pair is decomposed into five different terms giving you a wonderful 227255 numbers you have to visualize somehow. I haven't really figured out how to do this in a nice way, so instead I will plot the the total interaction energies between each unique pair of fragment for snapshot 2 and 3 for the blue curve (UCO) in the above graph.

I discovered the problem ... the fragmentation was not exactly the same along the entire UCO path which of course will make everything break down. Back to the drawing board then. What gave it away? Look here at the difference in pair-energies between frame 0 and frame 5 (click to see it in its full size - hell even that does not justify it)

So there it is. An unrealistic result can be your own fault no matter how hard you actually try to convince yourself that you used the same scripts for both runs.