[kriggy]

Hello guys, Im wondering if anyone has any experience with simulation behavior of molecules in silico? I would like to try to similate if my substituent will be bulky enough to block rotation around chirality axis. I think it shouldnt be that difficult but I never did or read anything at all about this kind of stuff. I have no idea where to start to look for something because there seems to be tons of different sofware available

Thank you
(feel free to move the topic I wasnt sure where to fit it)

[TheUnassuming]

I've never done energy calculations like that on my substrates, but others in my lab have.  I know the default program for a while in our lab was Spartan, but more recently I think people have been using Schrodinger.

[Enthalpy]

The first interrogation is: Linux or not Linux? Because about all this soft has been written for Linux. Only ports are available on Windows, with all the worries you imagine.

How complicated is your molecule? With 20 atoms, a PC can solve a molecular shape using things like UFF and AM1, but at 200 atoms the models are radically simplified and the solver differs.

There are many pieces of software, but once you notice that some are solvers and others are graphical interfaces, meant to work together on Linux, the choice narrows a lot.

[kriggy]

Im running on windows, the molecule is not realy complicated its something 18-30 atoms (including hydrogen) maybe - I know only the core structure  but I want to use the simulation to see what substituents might be worth tryting and what not

[Enthalpy]

I use the free ArgusLab, which installs cleanly on Windows and is decently fast to learn, while the competitors want you to compile a source meant for Linux or to rely on a binaries distribution that often targets the Win you don't have or need unincluded libraries like Gtk+. ArgusLab is also complete with the compute module AND the graphical interface, while most competitors are separated and need you to make some obscure link.

I've used regularly only that one, so I ignore what others may do better. Drawing a molecule is fast and almost intuive, ArgusLab optimizes the shape and claims to compute a (insufficiently accurate) heat of formation, and it displays the molecule nicely.

Runs on Windows 2000 and possibly older ones. Source:
http://www.arguslab.com/arguslab.com/ArgusLab.html

For you present need, maybe you could let optimize the shape, then rotate manually a bond to the worst angle, and let reevaluate the heat of formation without shape optimization. I suppose other programs are closer to this goal, but using ArgusLab this way may suffice to you. I ignore if ArgusLab lets freeze one angle or distance while optimizing the others, for instance allow a tBu to rotate when it's on the way of a C-C twist.

On my E8600 4GB, a molecule with 30 atoms gets its shape optimized with UFF in 10 seconds, then AM1 takes 10s more (appended example), good surprise. Molecule optimization doesn't use several cores nor the graphics card, but the 3D display does.

If you need only a few variations of one molecule, I could do it for you, but:

• Chances are that I misunderstand what you want;
• Once you have this possibility, you'll want to test many more groups and molecules;
• Having such a soft at hand is nice to represent molecules in a more realistic manner, for your needs and to show to other people as well,

so the present need can be a good opportunity to add ArgusLab (or an other) to your toolbox.

Chemists need the warning less than other people: simulation is not reality.

[kriggy]

Thank you, sadly for some reason argus stops responding when I run the shape optimalization

Thanks for the offer but I would prefer to do it on my own because It might be lots of structures and I would like to learn how to do this kind of stuff

[Enthalpy]

Learning Arguslab or an other is the reasonable option to my eyes too.

Stops responding: yes, it happens. I don't remember all the many causes for it. Some atoms aren't available for some Hamiltonians (like Br for Am1, sadly). Also, some Hamiltonians don't converge from too abnormal initial situations, especially excessive atom distances: Uff has the soundest behaviour and should be used first. A wrong number of electrons, say from mismatched sp² carbons, panicks the algorithm.

It helps the algorithm if the user constructs the molecule by steps, adding hydrogens automatically and converting them to groups, rather than placing each atom by hand.

Have a look at Arguslab's message window (mine is at the bottom right). For help calls like "Element 19 has no nshell" (= not modelled in Am1) or "Odd number of electrons" (= unbalanced sp² carbons), you can do something.

There are quite a few shortcomings in that soft that lets it behave a bit bizarrely, but over one or few days you learn to live with them in addition to learning the normal use . With the competitors, you first waste a good week of compilation and linking before you get the first message, so it's worse. Keep hope!