“…The best benchmarked modeling pair choices, as well as pairs which did not perform well, were considered for the analysis. The performance of the templates was ranked as good or bad, in published studies, on the basis of good ligand enrichment in VLS ( Perry et al, 2015 ; Loo et al, 2018 ; Jaiteh et al, 2020 ), local and global (RMSD) from crystal structures ( Castleman et al, 2019 ), and both ligand enrichment and RMSD from the crystal structure ( Shahaf et al, 2016 ). Researchers have compared varied parameters in these studies among the target-template pairs, including global sequence identity, TM-wise sequence identity, local sequence identity (identity within the binding pocket), model refinement through molecular dynamics and/or induced-fit docking, and the ligand binding site plasticity.…”