Balancing exploration and exploitation in population-based sampling improves fragment-based de novo protein structure prediction.

BackgroundFragment libraries play a key role in fragment-assembly based protein structure prediction, where protein fragments are assembled to form a complete three-dimensional structure. Rich and accurate structural information embedded in fragment libraries has not been systematically extracted and used beyond fragment assembly.MethodsTo better leverage the valuable structural information for protein structure prediction, we extracted seven types of structural information from fragment libraries. We broadened the usage of such structural information by transforming fragment libraries into protein-specific potentials for gradient-descent based protein folding and encoding fragment libraries as structural features for protein property prediction.ResultsFragment libraires improved the accuracy of protein folding and outperformed state-of-the-art algorithms with respect to predicted properties, such as torsion angles and inter-residue distances.ConclusionOur work implies that the rich structural information extracted from fragment libraries can complement sequence-derived features to help protein structure prediction.

One of the main research problems in Structural Bioinformatics is related to the prediction of three-dimensional structures (3-D) of polypeptides or proteins. The rate at which amino acid sequences are identified is increasing faster than the 3-D protein structure determination by experimental methods. Computational prediction methods have been developed during the last years, but the problem still remains challenging because of the complexity and high dimensionality of a protein conformational search space. In this article we present a hybrid genetic algorithm for the Protein Structure Prediction (PSP) Problem. A genetic algorithm is combined with a structured population, and it is hybridized with a path-relinking procedure that helps the algorithm to scape from local minima. We perform a set of experiments and show that the proposed hybrid genetic algorithm is effective in finding good quality solutions for the PSP Problem.

Most computational protein structure prediction methods are designed for either template based or template-free (ab initio) structure prediction. The approaches that integrate the prediction capabilities of both template-based modeling and template-free modeling are needed to synergistically combine the two kinds of methods to improve protein structure prediction. In this work, we develop a new method to integrate several protein structure prediction methods including our template-based MULTICOM server, our ab initio contact-based protein structure prediction method CONFOLD, our multi-template-based model generation tool MTMG, and locally installed external Rosetta, I-TASSER and RaptorX protein structure prediction tools to improve protein structure prediction of a fullspectrum difficulty ranging from easy, to medium and to hard. Our method participated in the 11(th) community-wide Critical Assessment of Techniques for Protein Structure Prediction (CASP11) in 2014 as MULTICOM-NOVEL server. It was ranked among top 10 methods for protein tertiary structure prediction according to the official CASP11 assessment, which demonstrates that integrating complementary modeling methods is useful for advancing protein structure prediction.

Protein structure prediction is one of the most challenging problems in computational biology. Given a protein's amino acid sequence, a simplified version of the problem is to find an on-lattice self-avoiding walk that minimizes the interaction energy among the amino acids. In this paper, we present a memory-based local search method for the simplified problem using Hydrophobic-Polar energy model and Face Centered Cubic lattice. By memorizing local minima and then avoiding their neighbohood, our approach significantly improves the state-of-the-art local search method for protein structure prediction on a set of standard benchmark proteins.

Local search can be introduced into genetic algorithms to create a hybrid, but any improvement in performance is dependent on the learning mechanism. In the Lamarckian model, a candidate solution is replaced by a fitter neighbour if one is found by local search. In the Baldwinian model, the original solution is retained but with an upgraded fitness if a fitter solution is found in the local search space. The effectiveness of using either model or a variable proportion of the two within a hybrid genetic algorithm is affected by the topology of the fitness function and the details of the hybrid algorithm. This paper investigates an intelligent adaptive approach to decide on the learning mechanism to be used by an individual over the course of the search. Evolution is used to self-adapt both the frequency of a steepest-descent local search and the relative proportions of Lamarckian and Baldwinian inheritance. Experiments have shown that this form of adaptive learning can improve the ability to find high-quality solutions and can accelerate the hybrid search without the need to find optimal control parameters for the learning process.

Operations performed in a Binary Search Tree generally starts from root node. As result search space (total number of nodes) constitutes entire tree, where a normal users intention is only certain part of the tree. In this paper we propose a noble method for performing operations such as insertion, deletion and retrieval within the local search space of a lookup node rather than the root node. To define local search space of a node, we implement leaf nodes null pointers, which are generally remains unused. These local search spaces divides the actual search space which generally constitutes entire tree. Hence complexity reduced to O(log m) from O(log n) for a local node, where m is the number of nodes present in sub tree formed by local search space and n is the number of nodes present in the tree.

A procedure has been developed for global energy minimization of surface loops of proteins in the presence of a fixed core. The ECEPP potential function has been modified to allow more accurate representations of hydrogen bond interactions and intrinsic torsional energies. A computationally efficient representation of hydration free energy has been introduced. A local minimization procedure has been developed that uses a cutoff distance, minimization with respect to subsets of degrees of freedom, analytical second derivatives, and distance constraints between rigid segments to achieve efficiency in applications to surface loops. Efficient procedures have been developed for deforming segments of the initial backbone structure and for removing overlaps. Global energy minimization of a surface loop is accomplished by generating a sequence (or a trajectory) of local minima, the component steps of which are generated by searching collections of local minima obtained by deforming seven‐residue segments of the surface loop. The search at each component step consists of the following calculations: (1) A large collection of backbone structures is generated by deforming a seven‐residue segment of the initial backbone structure. (2) A collection of low‐energy backbone structures is generated by applying local energy minimization to the resulting collection of backbone structures (interactions involving side chains that will be searched in this component step are not included in the energy). (3) One low‐energy side‐chain structure is generated for each of the resulting low‐energy backbone structures. (4) A collection of low‐energy local minima is generated by applying local energy minimization to the resulting collection of structures. (5) The local minimum with the lowest energy is retained as the next point of the trajectory. Applications of our global search procedure to surface segments of bovine pancreatic trypsin inhibitor (BPTI) and bovine trypsin suggest that component‐step searches are reasonably complete. The computational efficiency of component‐step searches is such that trajectories consisting of about 10 component steps are feasible using an FPS‐5200 array processor. Our procedure for global energy minimization of surface loops is being used to identify and correct problems with the potential function and to calculate protein structure using a combination of sequence homology and global energy minimization.

MEMPSODE: A global optimization software based on hybridization of population-based algorithms and local searches

The probability to predict correctly a protein structure can be enhanced through introduction of spatial constraints - either from NMR experiments or from homologous structures. However, the additional constraints lead often to new local energy minima and worse sampling efficiency in simulations. In this work, we present a new parallel tempering variant that alleviates the energy barriers resulting from spatial constraints and therefore yields to an enhanced sampling in structure prediction simulations.

Adoption of an improved PSO to explore a compound multi-objective energy function in protein structure prediction

In this paper, Protein Structure Prediction problem is solved using Diversity Controlled Self-Adaptive Differential Evolution with Local search (DCSaDE-LS). DCSaDE-LS, an improved version of Self-Adaptive Differential Evolution (SaDE), use simple fuzzy system to control the diversity of individuals and local search to maintain a balance between exploration and exploitation. DCSaDE-LS with four different local search replacement strategies are used. SaDE is also implemented for comparison purposes. Algorithms are tested on a peptide Met-enkephalin for force fields ECEPP/2, ECEPP/3 and CHARMM22. Results show that both DCSaDE-LS and SaDE produce the best energy for both force fields. Among the four replacement strategies, DCSaDE-LS1 strategy reports better results than other strategies and SaDE in terms of number of function evaluations, mean energy and success rate. Best conformations obtained using DCSaDE-LS is compared with native structure 1PLW and GEM structure Scheraga. Nonparametric statistical tests for multiple comparisons ($$1\times N$$1×N) with control method are implemented for CHARMM22 observations. A set of unique 100 best conformations obtained from DCSaDE-LS are clustered into 3 independent clusters suggesting the robustness of this methodology and the ability to explore the conformational space available and to populate the near native conformations.

Proteins are cellular macromolecules made up of linear chains of amino acids that adopt a unique three-dimensional structure to carry out specific biological functions. Amongst various computational approaches developed for Protein Structure Prediction (PSP), ab initio methods perform the prediction without prior knowledge of known structures. However, due to huge computational complexity incurred by these methods, simplified hydrophobic-polar (HP) models are often used for the investigations. Despite this simplicity, PSP continues to be a hard combinatorial optimization problem requiring development of efficient heuristic search techniques. Memetic algorithm (MA), with its flexible architecture, encompasses the strength of both local and global search and has shown promise in solving problems with complex search landscape, including PSP. This thesis is focused on developing an effective ab initio technique under a suitable MA framework for solving the PSP problem using low resolution HP models. To enhance the effectiveness of the MA, a mechanism is devised to include knowledge from the problem domain by exploiting the concept of hydrophobic-core (H-core) formation. The size of the maximal possible H-core is estimated and the approximation is validated by our proposed deterministic search technique. Further, a new knowledge-based initial population generation technique ensures generating conformations which are not only valid, but also maintain diversity in the population. This helps to commence the search with good seeds leading towards faster convergence. While solving the PSP problem using MA, the greatest intricacy arises from the complex nature of its multi-modal landscape. Like all the evolutionary algorithms (EAs), MAs have the tendency to get trapped into local minima due to selection pressure and genetic drift induced by the genetic operators, and techniques are necessary to surmount this difficulty. A novel parental selection technique called “Adaptive Strategy for Assortative Mating” is developed that dynamically distributes the re production opportunity among the individuals of the population to boost the effectiveness of cross-over. Further, a new survival selection strategy, namely “Sib-based Survival Selection”, prevents the rapid spread of genetic material from a particular individual to other members in the population, ensuring the concurrent preservation of several potential solutions, and restricting premature convergence. A new diversification mechanism, namely “Memory-based Diversification”, systematically diversifies the population by introducing new genetic material whenever it gets stuck at local peaks and allows it to proceed for further progress. An extensive experimental analysis is also reported on the influence of all three techniques proposed to maintain the diversity of population. Furthermore, in order to carry out the conformational search efficiently, especially with the increase in sequence length, the optimization technique is designed such that it provides an effective balance between exploration and exploitation. A robust Multimodal Memetic Framework has been developed where optimization is accomplished in hierarchical stages. The population of each stage is partitioned into three different states: an Exploratory state to locate promising peaks, an Exploitative state to fine-tune the solutions found in the Exploratory state of the previous stage, and finally, a Central state to carry out the formation of potential solutions to explore the undiscovered regions. Non-identical genetic operators have been implemented in each state to balance the exploration and exploitation by employing necessary selection pressure and diversity. A non-isomorphic encoding based technique has been developed to capture the sub-structures that are representing a particular region. Furthermore, a novel local search technique exploits already explored fit individuals by employing the knowledge of sub-structures. The efficacy of the proposed framework is demonstrated with an adequate number of distinctive data sets and comparison with state-of-the-art methods.

Chapter 22 - Real-Time Adaptive GPU Multiagent Path Planning

Difficulty in sampling large and complex conformational spaces remains a key limitation in fragment-based de novo prediction of protein structure. Our previous work has shown that even for small-to-medium-sized proteins, some current methods inadequately sample alternative structures. We have developed two new conformational sampling techniques, one employing a bilevel optimisation framework and the other employing iterated local search. We combine strategies of forced structural perturbation (where some fragment insertions are accepted regardless of their impact on scores) and greedy local optimisation, allowing greater exploration of the available conformational space. Comparisons against the Rosetta Abinitio method indicate that our protocols more frequently generate native-like predictions for many targets, even following the low-resolution phase, using a given set of fragment libraries. By contrasting results across two different fragment sets, we show that our methods are able to better take advantage of high-quality fragments. These improvements can also translate into more reliable identification of near-native structures in a simple clustering-based model selection procedure. We show that when fragment libraries are sufficiently well-constructed, improved breadth of exploration within runs improves prediction accuracy. Our results also suggest that in benchmarking scenarios, a total exclusion of fragments drawn from homologous templates can make performance differences between methods appear less pronounced.

Two current methods of global optimization are coupled to produce the Replica-Exchange method together with Monte Carlo-with-Minimization (REMCM). Its performance is compared with each separate component and with other global optimization techniques. REMCM was applied to search the conformational space of coarse grain protein systems described by the UNRES force field. The method consists of several noninteracting copies of Monte Carlo simulation, and minimization was used after every perturbation to enhance the sampling of low-energy conformations. REMCM was applied to five proteins of different topology, and the results were compared to those from other optimization methods, namely Monte Carlo-with-Minimization (MCM), Conformational Space Annealing (CSA), and Conformational Family Monte Carlo (CFMC). REMCM located global minima for four proteins faster and more consistently than either MCM or CFMC, and it converged faster than CSA for three of the five proteins tested. A performance comparison was also carried out between REMCM and the traditional Replica Exchange method (REM) for one protein, with REMCM showing a significant improvement. Moreover, because of its simplicity, REMCM was easy to implement, thereby offering an alternative to other global optimization methods used in protein structure prediction.