SSPT: Secondary Structure Prediction Triangle

In this paper, a novel technique, namely SSPT, is introduced to predict the secondary structure (SS) of a protein just based on its primary structure. In the training phase of this technique, a novel training tool (secondary structure triangle) is trained to reflect the tendency of overlapping small amino acid windows of a sequence toward three SS formation of H/E/L. These tendencies are then augmented to form three SS signals to reflect the neighboring properties of different sections of the sequence. These signals are then used to determine the protein´s class (alpha, beta, or alpha + beta) for better prediction of its structure. SSPT is tested using three well-known benchmarks (RS126, CB396, and CB513). Results are promising and authenticate the hypothesis behind this work.