Barbosa E. G., Martins J. P. A., Pasqualoto K. F. M., Ferreira M. M. C., "LQTAgrid: an open source package to generate 4D-QSAR descriptors". Porto de Galinhas, PE, Brazil, 09-13/11/2008: The 4th Brazilian Symposium on Medicinal Chemistry (BrazMedChem2008): Systems Chemical Biology, CD-ROM Online, (2008) No. 171.  Poster R0348-1.

Brazilian Chemical Society (SBQ). Division of Medicinal Chemistry. 4th Brazilian Symposium on Medicinal Chemistry

LQTAgrid: an open source package to generate 4D-QSAR descriptors

Barbosa, E.G.; Martins, J.P.A.; Pasqualoto, K. F. M.; Ferreira, M. M. C. - jpmartins@iqm.unicamp.br

Laboratório de Quimiometria Teórica e Aplicada - LQTA, Instituto de Química, UNICAMP, Campinas-SP, Brasil

Keywords: Molecular Dynamics, 4D-QSAR, Gromacs, LQTAgrid.
 

Introduction

3D-QSAR formalisms, such as comparative molecular field analysis (CoMFA),¹ use a set of compounds to generate 3D descriptors for building partial least squares (PLS) models, and provide relevant information for developing ligand-based drug design. Hopfinger and co-workers² reported an independent-receptor (IR) methodology where multiple conformations of each ligand obtained from molecular dynamics (MD) simulations are considered in the construction of IR 3D-QSAR models. Aiming to combine the advantages of both methods, CoMFA and IR 4D-QSAR, an open source package of programs was developed, named LQTAgrid.
Initially, the open source program GROMACS³ is employed to create a conformational profile (CP) of each ligand in the training set, from MD simulations having explicit solvent. Then, the CP of the ligands are aligned in a 3D virtual box or grid and the van der Waals and electrostatic energy contributions are calculated, using probes, to generate the 4D-QSAR descriptors matrix (LQTAgrid program). The construction of multivariate QSAR models can be performed according to the user's software preferences.
 

Results and Discussion

To validate the methodology proposed, the following two sets considering distinct classes were chosen: 44 inhibitors of p38 kinase⁴ (set 1) and 47 glucose analogue inhibitors of glycogen phosphorylase⁵ (set 2).
A previous variable selection was carried out, using the Pirouette package and the OPS algorithm⁶, which was developed in our research group. Reasonable QSAR models employing PLS and leave-one-out crossvalidation were obtained. The best QSAR model generated with set 1 presented the following statistical parameters values: q² = 0.70; r² = 0.83; and, standard error of calibration (SEC) of 0.26 and standard error of validation (SEV) of 0.30, with 3 latent variables (LV), which were statistically more significant than the values reported in ref. 4 [q² = 0.55; r² = 0.91; SEC = 0.19]. The values of q² and r² reported in ref. 4 are indicative of overffiting. Regarding set 2, the best QSAR model presented the values of statistical measures comparable to those from the original paper. LQTAgrid (q² = 0.76; r² = 0.80); ref. 5 (q² = 0.83 and r² = 0.87), SEV was 0.63 using 5 LV. Those QSAR models were validated applying Y-randomization and leave-N-out (N = 1 to 10) methodologies.
The descriptors selected in the best QSAR models can be graphically visualized (hot spots) in Figure 1. Favorable and unfavorable energy contributions (electrostatic and van der Waals) to the biological activity are defined based on the sign of the PLS regression coefficients. Those contributions correspond to possible ligand-receptor interactions, as well as favorable ligand occupations (places for adding functional groups that would increase the biological activity, for example). Figure 1 shows the graphical visualization of the 4D descriptors selected in the best QSAR model for a ligand from set 2.

Figure 1. Descriptors graphical representation considering the best QSAR model (set 2)
 

Conclusions

The methodology presented generates 4D descriptors, which after a variable selection, provide reliable and robust QSAR models. The collaborative license of the open source LQTAgrid program will allow its use for the construction of descriptor matrices in 4D-QSAR analyses.
 

Acknowledgements

The authors are grateful to CAPES, FAPESP and CNPq for the financial support.
 

¹ Cramer, R.D.; Patterson, D.E.; Bunce J.D. J. Am. Chem. Soc. 1988, 110, 5959;
² Albuquerque M.G.; Hopfinger, A.J.; Barreiro E. J.; de Alencastro R. B. J. Chem. Inf. Model. 1997, 38, 925;
³ Christen, M.; Hunenberger, P. H.; Bakowies, D.; Baron, R.; Burgi, R.; Geerke, D. P.; Heinz, T. N.; Kastenholz, M. A.; Krautler, V.; Oostenbrink, C.; Peter, C.; Trzesniak D. J. Comput. Chem. 2005, 26, 1719;
⁴ Ravindra, G.K.; Achaiah, G.; Sastry, G.N. Euro. J. Med. Chem. 2008, 43, 830.;
⁵ Venkatarangan, P.; Hopfinger, A. J. Chem. Inf. Comput. Sci. 1999, 39, 1141.
⁶ Teófilo, R.F.; Martins, J.P.A.; Ferreira, M.M.C. J. Chem. 2008, xx.
 
 
 
 

4th Brazilian Symposium on Medicinal Chemistry - BrazMedChem2008