Benchmark of 3D conformer generation and molecular property calculation for medium-sized molecules

Medium-sized molecules have attracted significant attention as new chemical modalities. In this study, we compared the performances of three methods 3D structure generation for medium-sized using free and commercial software. The benchmark dataset consisted 2131 protein-binding ligands with molecular weights greater than 600, which were selected from Protein Data Bank (PDB). When selecting smallest root mean square deviation between generated conformers PDB ligand structures, 43% conformations determined software CORINA within 1 Å, followed by 10% OMEGA 5% RDKit. According to our results, comparing polar solvent-accessible surface area (PSA) normalized principal moment inertia ratio (NPR) among methods, 83% 20 in PSA, 53% 0.05 NPR1 NPR2 spaces. Thus, concluded that has highest performance terms efficient conformer generation. We also examined descriptor calculation Mordred, is a computation tool. results showed OMEGA-generated exhibited success rate, indicating suitable tool various descriptors. Our could contribute selection conformation generators rapid construction predictive models can be shared research community further validation.