## Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

** enzyme modelling - entropy of a reaction and spin multiplicity **
alex

yakovenko.alexander@gmail.com

Dear QChemists,

I am modelling O2 molecule activation (from triplet to singlet) and the bond cleavage in oxidative enzyme as a function of different cofactors. For this purpose I generated a set of frozen O2 separations for 1.4-3.0A with a step of 0.1A (16 independent systems with frozen reaction coordinate O-O bond to 1.4, 1.5, 1.6, ... 3.0A), there is own set for each co-factor candidate. The surrounding protein environment consists of frozen root atoms bound to unconstrained side-chain fragments. I am optimizing the systems with the reaction coordinate constraint to obtain a reaction path (crossing fingers to obtain a tractable series of steps from reagents to the products). There are two questions about how to convert the series of intermediates into enzyme efficiency (as function of a cofactor).

1. Entropy of the reaction. As the first approximation a Boltzmann distribution can be calculated for a chance to pass the reaction barrier using discrete firefly energies. Is it possible/does it make sense to softly constrain the reaction coordinate and perform a sort of random walk around each rc value (like umbrella sampling technique). Then extract entropy with a (sort of) WHAM (weighted histogram analysis) approach. Is it worst to dig in for such second order approximation or it gives next to no benefits? If the answer is positive, can you give a hint of how to activate such random walk process in firefly and what statistics is so-so sufficient to extract the entropy?2. Spin calculations. Oxygen normally comes as triplet and turns into singlet form somewhere during the activation process. I suggesting finding of two minimums per each coordinate of reaction route: SCF=RHF, MULT=1 and SCF=ROHF,MULT=3 (or ROHF,1 and ROHF,3 is better?). Then use Boltzmann distribution to scan all possible routes (16 totally as the spin state can switch at any position) and choose the most ‘passable’ for the estimatiosn as described above. Does the scheme looks reasonable for experienced folk? Is there any easier/more accurate way to deal with the spin problem? Maybe there is some details to pay special attention?

Thanks,

Alex

Thu Aug 21 '14 10:15pm

This message read **687** times