Previous Up Next

Computer-Aided Drug Design Tutorials: 2. Computation of Descriptors for QSAR

QSAR

Quantitative Structure Activity Relationships are often used in ligand structure-based drug design. The QSAR relates drugs potency or toxicity with a variety of molecular descriptors. While many descriptors, such as the number of H-bond acceptors, are easy to calculate based on the molecular connectivity, other descriptors may require more advanced calculations. Notably, descriptors that may relate to chemical reactivity of drugs, such as HOMO and LUMO energies, must be obtained from quantum chemical calculation.

HOMO and LUMO energies

HOMO and LUMO refer to Highest Occupied Molecular Orbital and Lowest Unoccupied Molecular Orbital. According to the Frontier Orbital Theory, nucleophilic attack occurs by electron flow from a (HOMO of) nucleophile into the LUMO of the electrophile. In stable molecules, occupied electrons always reside on orbitals with negative energies and unoccupied orbitals have positive energies. The energies of HOMO and LUMO are related to the reactivity of the molecule: molecules with electrons at accessible (near-zero) HOMO level tend to be good nucleophiles because it does not cost much to donate these electrons toward making a new bond. Similarly, molecules with low LUMO energies tend to be good electrophiles because it does not cost much to place an electron into such orbital.

Example: Oxamic Acid

The Gaussian calculation you performed in the previous part produced description of molecular orbitals that MOLDEN is capable of visualizing. To visualize individual orbitals, click on Orbital while in the Density mode. Again, the Space mode (with contour value about 0.075) is easiest to grasp but Euclid is the fastest to quickly visualize orbitals. Notice that the first 6 orbitals in oxamic acid are strongly localized on atoms while many other orbitals are significantly delocalized. The orbital energies are also printed in the Gaussian output, under header Population Analysis Using the SCF Density.

For larger molecules, ab initio quantum mechanical calculations may become time consuming, especially because QSAR studies require that calculations are carried out on many molecules. Semiempirical calculations may be just accurate enough to give reasonably good descriptors for QSAR in a fraction of time. Many commercial programs, including Gaussian, and several free programs, such as MOPAC, and GAMESS, allow semiempirical calculations. Students who wish to try quantum chemistry calculations on their Windows or Linux PCs could download a free program PC GAMESS. Molden can create input files that (after some modification) are accepted by PC GAMESS outputs and MOLDEN can read some information from PC GAMESS output files. The last part of this tutorial illustrates how to perform semiempirical calculations with MOPAC through MOLDEN

  1. Read the structure of the cis conformer into MOLDEN by typing molden oxamic_ac_cis.mol2. The structure should show up in the graphical window.
  2. Open the Z-matrix editor but do not change the structure this time. Select Mopac from the Format menu and hit Submit Job.
  3. A new window, titled Submit Mopac Job, opens. Keep the Task to Geometry Optimization, and keep the Method AM1. By default, AM1 calculation uses a minimal fixed basis.
  4. Examine the molecule on the screen to verify that all formal valences are satisfied. Molecules like these are net neutral with all the electrons paired, hence keep 0 for charge and Singlet for Spin.
  5. Notice that the Mopac program uses keywords to decide what kind of calculation to perform and what kind of results to produce. The extra keywords, like GRAPH are required by MOLDEN to visualize results from Mopac. Add keyword MMOK to provide an additional correction to the AM1 description of the amide bond.
  6. Give an unique and easy-to-identify name to the job. For example, you may name it oxamic_ac_cis_am1.
  7. The title line is for optional comments. People typically add some description as of the purpose of this calculation. You may want to give your name here as it will help when your files are misplaced.
  8. Hit Submit, then OK. A confirmation window opens reporting that the job started. Hit OK to close this window.
  9. Unlike force field calculations, the results are not automatically returned to MOLDEN when the calculation completes. In fact, when running longer jobs, you may want to close MOLDEN, or even log out as the Gaussian calculation is now independent of MOLDEN. For oxamic acid, the calculation should take about 20-30 seconds to complete.
  10. There are two important output files. The file with gpt extension can be used by MOLDEN to visualize orbitals and electron density (but not the electrostatic potential). The file with out extension is human-readable and contains orbital energies near the end under EIGENVALUES header.
  11. Open a new connection to the workstation and examine the output file. Write down the HOMO and LUMO energy. To reuse the window in which MOLDEN was started, close molden, and type stty sane to reset the terminal.


Previous Up Next

Tutorial by Dr. Kalju Kahn, Department of Chemistry and Biochemistry, UC Santa Barbara. ©2005-2007.