In this study, the multivariate partial least squares projections to latent
structures (PLS) technique was used for modeling the RP-HPLC retention dat
a of 17 chalcones, which were determined with methanol-water mobile phases
of different compositions. The PLS model was based on molecular descriptors
which can be calculated for any compound utilizing only the knowledge of i
ts molecular structure. The PLS analysis resulted in a model with the follo
wing statistics: r = 0.976, Q = 0.933, s = 0.076, and F = 43.63. The adequa
cy of the developed model was assessed by means of cross-validation and als
o, by PLS modeling of the retention data of several chalcones reported by W
alczak et al. [J. Chromatogr. 353, 123, (1986)], which were obtained using
stationary phases of different polarity (-NH2, DIOL, -CN, ODS, C-8). The st
ructural interpretation of the developed PLS model was accomplished by mean
s of comparative correlations between the nonempirical descriptors used in
the model and the solvation parameters developed by Abraham. The results ob
tained in this work provides evidence for the great potential of the topolo
gical approach for the development of quantitative structure-retention rela
tionship (QSRR) models.