Article Preview
Top1. Introduction
Many essential functions for life are performed by proteins and the study of their structures yields the ability to elucidate their Functionalities and properties in terms of a molecular view (Creighton, 1992). In many fields, there is great interest in discovering a methodology for Protein Structure Prediction (PSP), including drug design, diseases mechanisms and food industry. In this context, some experimental methods have been applied to determine the structure of proteins, such as X-ray crystallography and nuclear magnetic resonance. Despite their success, both methods have characteristic limitations. Conversely, the knowledge of the primary sequence of the amino acids of a protein is a simpler experimental procedure.
In protein molecular modeling, the determination of the native spatial arrangement of the molecule atoms, which corresponds to a global or local energy minimum state, is a fundamental task because it represents the protein functional conformation. However, there is not an efficient general computational method to achieve this purpose yet (Leach, 2001). Indeed, one of the main braches of the protein folding area is the computational problem of how to predict a protein native conformation solely from its amino acids sequence (Dill et al., 2008).
Although it is not possible assure that the global free energy minimum always corresponds to the protein bio-active conformation, at least it will be a local minimum (Leach, 2001). Several computational methods for PSP are semi ab initio methodologies, so they also use prior knowledge from both the sequence homology and the statistics found on protein databases (Miyazawa & Jernigan, 1985; Poole & Ranganathan, 2006). However, the use of this additional information restricts the search of protein structures that could be correctly predicted from the vast universe of proteins.
Evolutionary Algorithms (EAs) have been investigated because they have capacity to exploit extensively the search space in order to look for the best solutions (Michalewicz and7 Schoenauer, 1996). EAs are a metaheuristics inspired by evolutionary theory (Goldberg, 1989). These algorithms have been applied in mono and multi-objective optimization problems (Talbi, 2009).
In fact, macromolecular modeling and design are increasingly useful in basic research, such as biotechnology. However, the absence of a user-friendly modeling framework that provides access to a wide range of modeling capabilities discourages the wider adoption of computational methods by non-experts (Fleishman et al., 2011). Moreover, PSP requires efficient computational methods (Nair and7 Goodman, 1998).
All frameworks provide some information about protein but not all of them. Therefore, it is necessary to work with more than one, although it is an uncomfortable situation because frameworks can be operating with different force fields and/or units. Furthermore, PSP is an open problem, where researchers have applied different ideas and methods. Therefore, developers are working to integrate these frameworks in order to improve their capabilities and make their use more straightforward (Fleishman et al., 2011; Cock et al., 2009; Lund et al., 2008; Eswar et al., 2002). In this way, ProtPred-GROMACS (Faccioli et al., 2011; Faccioli et al., 2012) has been applied in PSP as mono-objective concept. Here, it is presented in multi-objective concept through implementation of NSGA-II algorithm.