QSPR model for bioconcentration factors of nonpolar organiccompounds using molecular electronegativity distance vectordescriptors

摘要

A five-variable model (model M2) was devel-oped for the bioconcentration factors (BCFs) of nonpolarorganic compounds (NPOCs) by using molecular electro-negativity distance vector (MEDV) to characterize the struc-tures of NPOCs and variable selection and modeling basedon prediction (VSMP) to select the optimum descriptors. Theestimated correlation coefficient (r~2) and the leave-one-outcross-validation correlation coefficients (q~2) of model M2were 0.9271 and 0.9171, respectively. The model was exter-nally validated by splitting the whole data set into a repre-sentative training set of 85 chemicals and a validation set of29 chemicals. The results show that the main structural fac-tors influencing the BCFs of NPOCs are –cCc, cCcc, –Cl,and –Br (where "–" refers to a single bond and "c" refersto a conjugated bond). The quantitative structure—propertyrelationship (QSPR) model can effectively predict the BCFsof NPOCs, and the predictions of the model can also extendthe current BCF database of experimental values.