Protein Motif Retrieval by Secondary Structure Element Geometry and Biological Features Saliency

This paper presents an approach to detect the presence of a given motif in proteins or in protein data bank (PDB). The approach is based on the secondary structure elements (SSEs) geometrical arrangement in 3D space. A motif is represented as a set of SSEs in their specific positions related to a local reference system (LRS). We propose, exploiting the SSE biological feature saliency in the motif LRS construction stage, a planning strategy to speed-up the motif retrieval process. The experimentation has been carried out on a set of 20 proteins selected from the PDB. In detail we tested five different cases: (i) performances on searching a motif within single proteins, (ii) searching motifs on a set of proteins belonging to the same biological family, (iii) searching into single symmetric proteins, (iv) searching on a set of symmetric proteins from the same family, and finally (v) a general motif retrieval from the entire protein dataset. The experimental results showed good motif recognition performances on each test category, and, by exploiting the basic biological features saliency in motif construction, comparing to a previous approach of SSEs block geometrical retrieval based on the Generalized Hough Transform, it was revealed a significant decrease of the time/space computational complexity. It is worth to point out that the computation time for the case of motif absence is significantly lower than the case of motif present.