In
the present study, five detour/distance matrix based molecular
descriptors (MDs) termed as relative eccentric distance sum/product
indices (denoted by Rξ1
SV, Rξ2
SV, Rξ3
SV, RPξ1
SV, and RPξ2
SV), as well as their topochemical versions denoted by (Rξ1
cSV, Rξ2
cSV, Rξ3
cSV, RPξ1
cSV, and RPξ2
cSV) have been conceptualized
for exclusive use for molecules containing cyclic moieties. The said
MDs exhibited exceptionally high discriminating power and high sensitivity
toward branching/relative position of substituents in cyclic structures
amalgamated with negligible degeneracy. Subsequently, the proposed
MDs along with other MDs were successfully utilized for the development
of models for the prediction of human glutaminyl cyclase (hQC) inhibitory
activity using decision tree (DT), random forest (RF) and moving average
analysis (MAA). A data set comprising of 45 analogues of substituted
3-(1H-imidazol-1-yl) propyl thiourea derivatives
was used. DT identified proposed relative eccentric distance sum topochemical
index-1 as the most important MD. High accuracy of prediction up to
96%, 93%, and 95% was observed in case of models derived from decision
tree, random forest, and MAA, respectively. The statistical significance
of proposed models was assessed through specificity, sensitivity,
overall accuracy, Mathew’s correlation coefficient (MCC), and
intercorrelation analysis.