New Vertex Indices and their Applications in Evaluating Antileukemic Activity of 9‐Anilinoacridines and the Activity of 2′,3′‐Dideoxy‐Nuclosides Against HIV
Chandan Raychaudhury, Gilles Klopman · Bulletin des Sociétés Chimiques Belges · 1990
Abstract A newly developed graph‐theoretical artificial intelligence program has been used to evaluate qualitatively the antileukemic activity of 9‐anilino‐acridines and the activity of 2′,3′‐dideoxy‐nucleosides against HIV. The idea behind the approach is to predict the biological activity of any chemical compound on the basis of already identified pharmacologically/toxicologically relevant components of the molecules exhibiting the same kind of activity. A family of new graph‐theoretical (topological) vertex indices called distance exponent indices (Dx) of molecular graph have been proposed and used in the present study. Two previously used information‐theoretical topological indices viz., vertex distance complexity (Vd) and normalized vertex distance complexity (Vdn) have also been used. However, some distance exponent indices have produced the most significant results. These indices appear to identify biologically relevant molecular components (vertices in our case) quite effectively. Subsequently, an experiment of cross‐validation has been carried out to analyze the predictive power of the present approach. The possible application of the vertex indices, considered here, in different databases has also been discussed.