Original scientific paper

A Strategy for Molecular Modeling of a Physicochemical Property using a Linear Combination of Connectivity Indexes

Lionello Pogliani ; Dipartimento di Chimica, Università della Calabria, 87030 Rende (CS) Italy

Abstract

A strategy to employ a linear combination of the connectivity indexes (LCCI) method to model a physicochemical property of a series of molecules is outlined throughout. The chosen physicochemical property is the solubility of 19 natural amino acids. This property is modeled with the aid of normal LCCI and of linear combinations of special constructions of connectivty indexes relating to different numbers of amino acids, including and excluding extreme outliers. The employed indexes are analyzed following their descriptive power of solubility of natural amino acids. A linear combination of reciprocal connectivity indexes (LCRCI) showed the best mapping of the water solubility of 16 amino acids while a model based on LCRCI, along with the use of supraconnectivity indexes for the amino acids proline, serine and arginine, achieved very good modeling of the water solubility of the entire set of n = 19 natural amino acids. Linear combinations of fragment connectivty indexes were also analyzed while linear combinations of orthogonal connectivty indexes (LCOCI) were used to improve the modeling and to detect dominant descriptors for this physicochemical property. Non connectivity 'ad hoc indexes, already used in a previous study with good success, did not achieve the same good quality as the linear combination of reciprocal connectivity indexes.

Keywords

Hrčak ID:

135948

URI

https://hrcak.srce.hr/135948

Publication date:

4.3.1996.

Visits: 640 *