# Why are MD (Molecular Dynamics) simulations performed at room temperature for the purposes of studying proteins?

All of the MD simulations which are cited in the literature seem to be performed at room temperature (~300 Kelvins), while enzymes are usually bioactive at body temperature (~310 Kelvins).

I am wondering why not just use 310 K for the MD simulation? I suspect it is because most water models are designed specifically for room temperature (ie. TIP3P waters).

• Great question. I coauthored some MD studies between 2000 and 2006, where we looked at motor proteins (ncd, myosin, kinesin) and sure enough, we went with 300 K. In many cases, you're not dealing with a human system - myosin from $\it{Dictyostelium\,discoidium}$, ncd from $\it{Drosophila}$ - so human body temperature is not appropriate. Also, at least in the studies I was involved in, you're interested at normalizing computational conditions to experimental, which many times turn out to be at room temperature. – Todd Minehardt Jan 1 '18 at 21:33
• I'm no MD expert, but I don't think a reliable water model at 300K would be bad at 310? – Karl Jan 3 '18 at 13:06

Another reason why people don't bother to use 310K versus 300K is because any time you are using MD, you have an approximate force field describing the dynamics of the system which means you have an approximate phase diagram for the system. When comparing MD and experiment, it is a reasonable question if you really want to compare properties at the same temperature or compare properties at "equivalent" points on the phase diagram. As an example, some models of water which are very widely used have a phase diagram which is drastically different from the real phase diagram of water (i.e. they boil at 400K and other weird things) and you will see studies which describe the phase diagram of DFT water or TIP4P water so that you can make comparisons based on the temperature and on the phase diagram.

So, this is to say, if you submit a paper where you do MD at 300K when studying an enzyme in solution, it is unlikely you will get complaints that this is a biomolecule so it should be at 310K because even comparing to properties of the actual enzyme at 310K isn't a direct comparison since you aren't at the same locations on the respective phase diagrams. What could be more relevant is to make comparisons to the real world across a range of temperatures so that you can see if the trends are the same rather than just a single point.

Another less physical reason is that people have already been doing simulations at 300K, so you can't really get complaints about a simulation at 300K since everyone has already been using 300K. This is actually very common in science.

This depends on the organism you are studying. I work with Bac and human, I have simulations running at 300k and 310k. Search on the organism you are going to work on and adjust its temperament for your simulation.

You want to be able to compare your simulation with the real world, and that means you do (or look at other's) in vitro experiments, which are predominantly (at least for starters) done at room temperature because that's simplest. Further you want to be able to compare to results of earlier simulations (other MD engine, varied chemical system), and you want at least the same (nominal) temperature and pressure in your simulation. 300K is the standard, so you stick to that.

Also who says your enzymes(, ... ) stem from warm-blooded animals?

I second jheindel's suggestion that people tend to just follow the literature. It's also true that the fluctuation in temperature during your MD simulation is consistently on the range of several kelvin. Therefore, 310 and 300 K aren't that far apart.

Nonetheless, the conformational space sampled by the protein is heavily dependent on temperature (think of denaturation temperatures as an extreme example). As such, we should try our best to replicate physiological conditions. If the optimal growth temperature or physiological temperature of the biological host from which you took your protein is known, you should run your simulations at that temperature. This is especially relevant for extremophilic proteins, such as thermophiles and cryophiles.

• Are your forcefields good enough to capture 10K difference? If not, it makes sense to perform most calculations at a standard condition, since it is easier to compare them with others, except if you do the study on temperature dependence (where I do not think you can rely on the absolute scale). Also, biochemistry is not dramatically different at room temperature from body temperature, and warm-blooded animals have essentially the same stuff as any other organism. Much lower or higher, sure, that makes difference, but again, we talk about 10 K. – Greg Mar 7 '19 at 5:21
• See Table 3 here (pubs.acs.org/doi/pdf/10.1021/jacs.7b11926) where a 17 K difference (using pretty standard Amber12 FF and TIP3P models) led to a different mechanism of electron transfer for E. Coli. – jezzo Mar 7 '19 at 20:11