The QSAR and docking studies were performed on fifty seven steroids with binding affinities for
corticosteroid-binding globulin (CBG) and eighty four steroids with binding affinities for sex hormone-binding globulin
(SHBG). Since the steroidal compounds have binding affinity for both CBG and SHBG, multi-target QSAR approach was
employed to establish a unique QSAR method for simultaneous evaluation of the CBG and SHBG binding affinities. The
constitutional, geometrical, physico-chemical and electronic descriptors were computed for the examined structures by
use of the Chem3D Ultra 7.0.0, the Dragon 6.0, the MOPAC2009, and the Chemical Descriptors Library (CDL) program.
Partial least squares regression (PLSR) has been applied for selection of the most relevant molecular descriptors and
QSAR models building. The QSAR (SHGB) model, QSAR model (CBG), and multi-target QSAR model (CBG, SHBG)
were created. The multi-target QSAR model (CBG and SHBG) was found to be more effective in describing the CBG and
SHBG affinity of steroids in comparison to the one target models (QSAR (SHGB) model, QSAR model (CBG)). The
multi-target QSAR study indicated the importance of the electronic descriptor (Mor16v), steric/symmetry descriptors
(Eig06_EA(ed)), 2D autocorrelation descriptor (GATS4m), distance distribution descriptor (RDF045m), and atom type
fingerprint descriptor (CDL-ATFP 253) in describing the CBG and SHBG affinity of steroidal compounds. Results of the
created multi-target QSAR model were in accordance with the performed docking studies. The theoretical study defined
physicochemical, electronic and structural requirements for selective and effective binding of steroids to the CBG and
SHBG active sites.