1. Computational Chemistry A teaser introduction José R. Valverde CNB/CSIC © José R. Valverde, 2013 CC-BY-NC-SA

2. Introduction Computational chemistry is a branch of chemistry that uses principles of computer science to assist in solving chemical problems. Taken from Wikipedia (don't forget your contribution).

3. VirtualBox We will use a virtual machine to demonstrate the principles involved. To run this virtual machine you need to install VirtualBox (http//www.virtualbox.org) in your computer. A virtual machine is just like a separate computer but with no real hardware: it borrows the hardware from your computer --while you are still doing other things.

4. Open VirtualBox and Start the machine

5. When the machine ends booting you will be faced by a login screen. Use password “freebit” to log in. CAUTION: THIS IS HIGHLY UNSAFE AND YOU SHOULD NEVER EVER USE SUCH AN OBVIOUS PASSWORD IN REAL LIFE. The first thing you should do in such a situation would be to change the password immediately to something safer.

6. Start Gabedit http://gabedit.sourceforge.net We will use Gabedit in this session. Gabedit is freely available for Mac, Linux and Windows.

7. Gabedit main window Gabedit relies on other programs to do the calculations. Many of these other pograms are also free as well (MPQC, NWCHEM, etc...)

8. Getting started To begin, we need a molecule to work with. Click on the small icon with a small molecule (labelled “Draw a Geometry” to get started.

9. Load a PDB molecule Instead of drawing a molecule, we will just load a small one, aspirin, to save up time. Press the right mouse button in the drawing window to bring up the menu and select “Read a PDB file”.

10. Open a molecule Normally you'd get it from a database (like PDB) but we will use a file already saved. Note that navigating the directories/folders is easy. And note as well that recently used folders are remembered.

11. Set atom types While not strictly needed for many tasks, this is very important. Not all atoms are born equal. Context may be very important even for the same kind of atom. We should always assign atom types first.

12. Molecular Mechanics MM treats atoms as charged soft balls and bonds as springs. While not chemically accurate it is fast!. Let us start by optimizing the structure (look for its lowest energy conformation). You can do this from the menu.

13. MM Gradient options This is nothing but a mathematical optimization problem. The most commonly used optimization methods are available and you can fine tune their parameters.

14. Optimized structure All we need to do is be patient. After a short while, we should have an “optimal” structure.

15. Minimization This is a purely mathematical problem. optimize potential energy (interatomic interactions) But molecular functions have specific features: they are multidimensional they yield complex hyperplanes It is possible that the optimization method ends finding a local minimum in the potential energy surface Ironsmithers found a clever way to make good swords very long ago

16. Conformational Search We can compute interaction forces and from them: -velocities -changes with time and energy (temperature) Heating we can exit local wells and search for better conformations.

17. Conformational search We will heat the molecule and record the 10 best conformations found. We will also further optimize each of them. Remember to select where to save the results so you can find them later.

18. Simulated annealing We are interested in heating the molecule to a large temperature to help it escape local minima. You can save the intermediate conformations (trajectory) and calculated properties as well as set a number of parameters.

19. Gradient options Remember, we are looking for the best conformation We also need to state which mathematical approach shall be used to detect minima.

20. The force field This is just an esoteric name for the set of parameters we use to define interatomic interactions. Amber is likely the most popular one. It defines atom radius, bond flexibility, charges, etc...

21. A general potential force field Potential energy is a function of atomic positions V = E bonded + E non-bonded E bonded = E bond-stretch + E bond-bend + E bond-rotation

22. Non-bonded interactions Sum of all inter-atomic interactions Van der Waals Electrostatic Hydrogen bonding (optional) Must be computed for each atom against all others (N 2 ) Van der Waals Computed as a Lennard-Jones potential Electrostatic Computed as classical interaction

23. Heating the molecule As you can see, the conformational space explored includes all sorts of conformations. The best 10 will be saved.

24. Final optimization After the heating is finished, each of the saved conformations is successively optimized in turn.

25. Final geometries Final results are saved for you to review.

26. Modern Chemistry With increased computer power, interest has raised on using more accurate methods. Quantum mechanics considers all electrons and so can provide a better description, but can easily become prohibitively expensive. A middle ground -semiempirical methods- substitutes as many QM parameters as possible by experimental data, speeding up calculations significantly.

27. MOPAC Optimization Click the right mouse button to bring up the menu and select Semi- empirical optimization using Mopac. Note that you can use other programs too.

28. MOPAC Options You can now select the quantum parameters to use. Note that you can add additional keywords to fine tune the calculation (e.g. adding solvent) and do not forget to select the working folder and output file.

29. Monitor optimization results It is always wise to check how did the calculation go You can see each tested conformation clicking on the dots in the energy graph.

30. SE conformational search Just like with MM we might have found a local minimum. We can also heat and seek.

31. SE General options Select the number of local minima (best conformers) to keep, where to save the results and whether each conformer should be further optimized and to which extent.

32. SE Dynamics options We'll select a short simulation for the tutorial. Quantum mechanics is much more expensive.

33. Model We no longer need a force field, but must state the quantum model to use. You can fine tune it if you want (e.g. to add solvent)

34. Explore the results Once done, we can explore the conformers selected and their relative energies clicking on the energy graph.

35. Molecular Dynamics Once we know we have a good starting structure we can now consider studying its dynamical properties. Typically, we will start by heating to room temperature, then allow the molecule some time to stabilize at the target temperature before starting the collection of data (production). Often we will also want to cool down the structure at the end (do a final optimization) to see if it returns to the original structure (ever fried an egg?)

36. Molecular Dynamics

37. MD parameters As you can see, the times for heating, equilibration... have changed. So has the temperature which is more sensible (300K approximates room temp.).

38. Why MD Running MD simulations yields lots of valuable information: Average values average conformation, viscosity, movility, charge orientation, stability... Dynamic values flexibility, movement, changes with different conditions (T, E, V...), intermolecular interactions, etc,.. Point values (maxima, minima, extremes...)

39. MD limitations Can aspirin stand 1000K? Likely not. MD does not allow for bond-breaking/formation MD cannot be used to model chemical interactions MD cannot explain reactivity or electronic properties But CAN provide very useful insights Atoms DO NOT exist in molecules. the electron cloud spreads around in all the space concentrating in areas close to nuclei.

40. MOPAC Click on the MOPAC icon to start a new semi empirical quantum mechanics calculation http://www.openmopac.net

41. Single-Point calculation A single-point calculation only computes the electronic distribution in the defined state. ergoscf.org

42. MOPAC input file You can modify the MOPAC input file to adapt it to your needs.

43. Save file Save the file you created so that you can find it later.

44. Save file dialog Try to remember the name you give the file. A good name would indicate the compound, the calculation done and its order in your work flow.

45. Run Click on the small clockwork icon to run a program.

46. Running MOPAC Verify that the program to be run is the one you want.

47. Output tab Select the output tab by MOPAC input file to see the results of the calculation. Note that it may take long to finish and net be terminated yet.

48. Update/end Click on “Update/end” to re-read the output file.

49. NWchem Click now the NWchem icon to start a new calculation using a more accurate method (a fully ab initio model). http://www.nwchem-sw.org

50. Do a single-point calculation Use Pople basis set 3-21G. This is not too accurate but runs fast enough for a first approximation. Normally we would use more accurate (and slower) basis sets afterwards.

51. Run program from menu It's the same

52. Run parameters Note that you can run programs on remote (and likely more powerful) computers.

53. Go to the output tab

54. Press Update/End This will take longer, you will need to keep updating until the program finishes.

55. Wait Ab initio calculations can take a very long time. You know when they end by looking at the end of the output.

56. You may want to try... MPQC http://mpqc.org GAMESS-US http://www.msg.ameslab.gov Firefly/PC-GAMESS http://classic.chem.msu.su ABINIT http://abinit.org PSI4 http://www.psicode.org FreeON http://freeon.org ErgoSCF http://ergoscf.org SIESTA http://www.icmab.es/siesta ORCA, Octopus, Quantum Espresso, ACES3...

57. Go back to MOPAC run

58. Visualize orbitals/density Click on Dens/Orb. A new window opens. There use the right mouse button to open a mopac AUX file from the menu.

59. Select molecular orbital You'll get a listing of molecular orbitals. Select the one you want to visualize (Occ. means occupancy).

60. Select calculation precision

61. Define isovalue for surface We cut the MO surface at a given density value which corresponds to a probability of finding the electron.

62. Aspirin orbital

63. Select a different MO

64. Choose LUMO LUMO is the first MO with an Occ. value of zero

65. Aspirin LUMO

66. Open NWchem output

67. Select orbital Note that with NWchem output the HOMO is not automatically selected.

68. Select HOMO HOMO is the last orbital with an Occ. value of one.

69. Aspirin HOMO

70. Electronic density

71. Aspirin electronic density

72. Chemical propensity

73. Some times you see nothing

74. Change surface isovalue

75. Automatically find isovalue

76. Aspirin Fukui calculation

77. Visualize MD trajectory

78. Open a saved trajectory

79. Play MD trajectory

81. Thanks To all of you For coming... and not falling asleep To the organizers For this wonderful opportunity To CNB/CSIC, EU-COST, CYTED For funding

82. Questions?