Abstract
The solvent accessibility of charges is introduced in QSAR studies. Atomic charges which do have contribution to the biological activity of a compound are more or less exposed to the environment (solvent), i.e. they have different possibilities to interact with biological macromolecules. Although one can imagine different ways in considering solvent accessibility, in this work it is taken into account by multiplying the net charge of an atom with its solvent accessible surface. To study the effect of accessibility of a charge on a biological activity, a large set (of 101) HEPT-based HIV- 1 reverse transcriptase (RT) inhibitors is analyzed, a molecu1a.r system, where the importance of electronic contribution is known. The results, based on multiple linear regression (MLR) and partial least squares (PLS) analysis, show that accessible charges correlate better with the RT inhibitory activity than net charges, leading to models with higher predictive ability.