A GPU-based hybrid jDE algorithm applied to the 3D-AB protein structure prediction

摘要: Abstract The Protein Structure Prediction (PSP) problem is one of the most significant open problems in bioinformatics. In AB off-lattice model, protein sequence labeled as ‘A’ or ‘B’ according to amino acid classification being hydrophobic hydrophilic. It has been widely explored literature because polarity main driving forces behind structure definition. This work provides a high-performance hybrid algorithm approach 3D-AB model through Graphics Processing Units (GPUs). proposed algorithm, named cuHjDE–3D, self-adaptive Differential Evolution (DE) that uses jDE mechanism self-adapt DE parameters and employs Hooke-Jeeves Direct Search (HJDS) exploitation routine. experiments were conducted on real sequences from Data Bank (PDB) compared against state-of-the-art algorithms related concerning model. Moreover, we provide methodology compare predicted conformation with its native PDB repository using RMSD metric. obtained results highlight optimization potential method. Also, GPU running time analysis reports positive impact massively parallel architecture, speedups up 277× , promoting necessary scalability handle