Comparison of Monte Carlo simulations of cytochrome b6f with experiment using Latin hypercube sampling

We have programmed a Monte Carlo simulation of the Q-cycle model of electron transport in cytochrome b(6)f complex, an enzyme in the photosynthetic pathway that converts sunlight into biologically useful forms of chemical energy. Results were compared with published experiments of Kramer and Crofts (Biochim. Biophys. Acta 1183:72-84, 1993). Rates for the simulation were optimized by constructing large numbers of parameter sets using Latin hypercube sampling and selecting those that gave the minimum mean square deviation from experiment. Multiple copies of the simulation program were run in parallel on a Beowulf cluster. We found that Latin hypercube sampling works well as a method for approximately optimizing very noisy objective functions of 15 or 22 variables. Further, the simplified Q-cycle model can reproduce experimental results in the presence or absence of a quinone reductase (Q(i)) site inhibitor without invoking ad hoc side-reactions.