[chysce]

So... I'm doing some basic physicochemical predictions like pKa, logP for my compounds (depsipeptides). Which is I guess fine to do with Marvin.. but... I'm also trying to make some cyclic depsipeptides so I was wondering if some calculations of linear depsipeptide conformation in Marvin would be reliable enough to see whether the molecule has a suitable conformation for cyclization or not? And if it is... could such thing be mentioned in a scientific paper?

[BobfromNC]

It is as accurate as most property predictions, which are just as good a election polling results, predicting how long you will live, picking good stocks, and the horse race handicapping papers.    The simple properties are fine, but almost all conformational predicts exclude the effects of solvent, and simplify most other factors, so they are not that good without some tweaking for most advanced work.   If you have similar molecules with known data and can test them first, that may tell you if the model is close on your molecules.  But they are always predictions and not real data, so take them with a grain of salt.