Scores of extended connectivity fingerprint as descriptors in QSPR study of melting point and aqueous solubility

QSPR studies, using scores of SciTegic's extended connectivity fingerprint as raw descriptors, were extended to the prediction of melting points and aqueous solubility of organic compounds. Robust partial least-squares models were developed that perform as well as the best published QSPR models for structurally diverse organic compounds. Satisfactory performance of the QSPR models indicates that the scores of extended connectivity fingerprint are high performance molecular descriptors for QSAR/QSPR studies. Performance of the fingerprint-based descriptors is further validated by the satisfactory prediction of aqueous solubility of nearly 1300 organic compounds (squared correlation coefficient of 0.83 and RMSE of 0.85 log unit) with Yalkowsky's general solubility equation using both calculated melting points and calculated octanol-water partition coefficients. It demonstrates for the first time that it is feasible to predict aqueous solubility of structurally diverse organic compounds with the general solubility equation using both the calculated melting points and the partition coefficients.