Abstract
A series of CoMFA models have been derived from docking-based and atom-based alignments. The statistics of these approaches has been compared to determine whether a docking approach can be employed as an automated alignment tool for the development of 3D-QSAR models. Using a well-characterized training set of 51 biphenyl carboxylic acid MMP-3 inhibitors, the docking-based alignment provided by a DOCK4/PMF-scoring protocol has yielded statistically significant, cross-validated CoMFA models comparable to those derived with a traditional atom-based alignment technique. Field fit minimization has been applied to refine the atom-based and docking-based alignments. The refinement appears to be beneficial for the docking-based approach. For the atom-based alignment, however, field-fit refinement has not resulted in improved CoMFA models. The statistically best CoMFA model has been created by the atom-based alignment that has been found, however, to be inconsistent with the stromelysin crystal structure. The docking alignment refined by field-fit alignment has resulted in a final alignment that is consistent with the crystal structure and only slightly statistically inferior to the atom-based aligned CoMFA model. The results show␣the ability of an automated docking/field-fit alignment technique to provide self-consistent CoMFA alignments.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have