- Title
- An improved memetic approach for protein structure prediction incorporating maximal hydrophobic core estimation concept
- Creator
- Nazmul, Rumana; Chetty, Madhu; Chowdhury, Ahsan
- Date
- 2021
- Type
- Text; Journal article
- Identifier
- http://researchonline.federation.edu.au/vital/access/HandleResolver/1959.17/176647
- Identifier
- vital:15156
- Identifier
-
https://doi.org/10.1016/j.knosys.2018.06.022
- Identifier
- ISBN:0950-7051 (ISSN)
- Abstract
- Protein Structure Prediction (PSP) from the primary amino acid sequence, even using a simplified Hydrophobic-Polar (HP) lattice model, continues to be extremely challenging. Finding an optimal conformation, even for a small sequence, by any of the currently known evolutionary approaches is computationally extensive and time consuming. Although Memetic Algorithms (MAs) have shown success in finding the optimal solution for PSP, no significant work on the incorporation of domain or problem specific knowledge into the search process to significantly improve their performance is reported. In this paper, we present an approach to incorporate such knowledge into the initial population to enhance the effectiveness of MA for PSP. The domain knowledge we propose to use is based on the concept of maximal ‘core’ formation by exploiting the fundamental property of the H residues to be at the core of the minimum energy optimal protein structure. A generic technique is proposed for estimating the maximal Hydrophobic core (H-core) in a protein sequence for 2D Square, 3D Cubic and a more complex and realistic 3D FCC (Face Centered Cubic) lattice models. Subsequently, the knowledge of this estimated core is incorporated in an MA. The experiments conducted using HP benchmark sequences for 2D Square, 3D Cubic and 3D FCC lattice models show that the proposed MA with the new core-based population initialization technique has superior performance to the existing methods in terms of convergence speed as well as minimal energy. © 2018 Elsevier B.V.
- Publisher
- Elsevier B.V.
- Relation
- Knowledge-Based Systems Vol. 219, no. (2021), p. 104395
- Rights
- All metadata describing materials held in, or linked to, the repository is freely available under a CC0 licence
- Rights
- Copyright @ 2018 Elsevier B.V. All rights reserved.
- Subject
- 08 Information and Computing Sciences; 15 Commerce, Management, Tourism and Services; 17 Psychology and Cognitive Sciences; Hydrophobic core; Lattice model; Memetic algorithm; Protein structure prediction
- Reviewed
- Hits: 1832
- Visitors: 1778
- Downloads: 0