The apparent Caco-2 permeability coefficients for 100 structurally diverse compounds were taken from 13 literatures. The data set was splited into a training set of 77 compounds and a test set of 23 compounds. The structures of the compounds were built within SYBYL 6.9 molecular simulation package and modeled in their neutral forms. Geometry optimization was performed by using molecular mechanism. In order to decrease the deviation produced by different assays, the caco-2 values averaged from different experimental sources were used in the current work. The molecular name, experimental logD, radium of gyration (rgyr), high-charged polar surface area (HCPSA), and fraction of rotatable bond (fROTB) are also included in the database files.

Download the databases:

training_set.sdf(the SDF file includes 77 molecules)
test_set.sdf(the SDF file includes 23 molecules)

Reference
[1]. Tingjun Hou, Wei Zhang, Ke Xia, Xiaojie Xu, ADME evaluation in drug discovery. 5. Correlation of Caco-2 permeation with simple molecular properties, Journal of Chemical Information and Computer Sciences , 2004, 44, 1585-1600. [html] [PDF]