Topological Indices and Graph Theory: a Useful Tool

TALEVI ALAN; BELLERA CAROLINA L.; BRUNO-BLANCH LUIS E.

Rio de Janeiro, Brasil

Simposio; Brazilean Symposium on Bioinformatics 2007; 2007

Pontificial Catholic University/Rio de Janeiro

Recently, interest has grown in the prediction of properties and function of peptides and proteins, arising the need to identify molecular descriptors suitable for establishing structure-property relationships. In order to prove if topological indices derived from Graph Theory could be applied in the characterization of peptides, we have generated two sets of peptides: one composed by the 120 possible pentapeptides obtained from permutations of five amino acids; the other composed by 30 octapeptides obtained from permutations of eight amino acids. Dragon software was applied in the calculation of 578 topological descriptors. The range of values assumed by each descriptor, its information content (quantified through Shannon Entropy) and degree of degeneracy are proposed as criteria for the selection of the descriptors which best characterize the peptides sets. Results indicate that topological indices may be valuable in the design of peptides with therapeutic applications and, in time, for the characterization of proteins.