The performed research focused on analyzing the herbicidal properties of several
new analogs of 6-(5-aryl-substituted-1-pyrazolyl)-2-picolinic acid using three
different methods, namely partial least squares, PLS analysis, the combinatorial
protocol in multiple linear regression, CP-MLR and the non-parametric Fujita-
Ban, FB technique. The FB method employs the principle of additivity, which
suggests that substituents contribute to the activity of the parent molecule. The
study investigated the impact of different substituents, positioned in various
ways, on the activity of the analogs. The most active compound in the series
exhibited a pattern of substituents exhibiting positive contribution relative to the
parent molecule. The CP-MLR technique employs statistically derived
quantitative structure-activity relationship, QSAR models to elucidate the
herbicidal action of congeners. In this procedure, three-dimensional descriptors
such as asphericity, ASP, the second component accessibility directional WHIM
index weighted by atomic masses, E2m, and the maximum autocorrelation of lag
2 weighted by atomic masses, R2m+ were identified as the most important.
Furthermore, external validation of the models was performed using data
generated from the test-set, and the models demonstrated their capacity for
forecasting through applicability domain, AD study. These findings can be
helpful in identifying potential analogs of the series. The dominance of the
descriptors discovered by the CP-MLR investigation was corroborated by PLS
analysis, and the computed activity values were shown to agree with observed
ones using the FB, CP-MLR, and the PLS analyses.
Keywords: QSAR study, Molecular 3D-descriptors, Herbicidal activity,
Derivatives of 6-(5-aryl-substituted-1-pyrazolyl)-2-picolinic acid