Optimal graph-based and simplified molecular input line entry system-based descriptors for quantitative structure-activity relationship analysis of arylalkylaminoalcohols, arylalkenylamines, and arylalkylamines as σ₁ receptor ligands

In this investigation, an optimal description of the structure-activity relationship was done for 59 σ₁ receptor ligands including arylalkylaminoalcohols, arylalkenylamines, and arylalkylamines (denoted as RC-33 analogs in this manuscript). We used optimal graph-based (GBD) and Simplified Molecular Input Line Entry System (SMILES)-based (SBD) descriptors as molecular characteristics of the structures to explain the differences in σ₁ receptor affinity between the ligands. The best graph-based descriptor model (named HFG-EC0 in the manuscript) included hydrogen-filled molecular graphs using the extended connectivity of zero order (EC0). This model has an average correlation coefficient value for external test set of 0.786 with a better statistics when comparing with the best SBD model. The best SBD model (named SSS_k in the manuscript) includes only three SMILES elements SSS_k and has an average correlation coefficient value for external test set of 0.726. These models identified the molecular features that contribute to a high σ₁ receptor ligand affinity.