Coevolutionary insights between promoters and transcription factors in the plant and animal kingdoms.

The divergence and continuous evolution of plants and animals contribute to ecological diversity. Promoters and transcription factors (TFs) are key determinants of gene regulation and transcription throughout life. However, the evolutionary trajectories and relationships of promoters and TFs are still poorly understood. Here, we conducted extensive analysis of large-scale multi-omics sequences in 420 animal species and 223 plant species spanning nearly a billion years of evolutionary history. Results showed that promoter GC-content and TF isoelectric points, as features/signatures that accompany long biological evolution, exhibited increasing growth in animal cells but a decreasing trend in plant cells. Furthermore, the evolutionary trajectories of promoter and TF signatures in the animal kingdom provided further evidence that Mammalia as well as Aves evolved directly from the ancestor Reptilia. The strong correlation between promoter and TF signatures indicates that promoters and TFs formed antagonistic coevolution in the animal kingdom, but mutualistic coevolution in the plant kingdom. The distinct coevolutionary patterns potentially drive the plant-animal divergence，divergent evolution and ecological diversity.

Loop Enhanced Conformational Resampling Method for Protein Structure Prediction.

Protein structure prediction has been a long-standing problem for the past decades. In particular, the loop region structure remains an obstacle in forming an accurate protein tertiary structure because of its flexibility. In this study, Rama torsion angle and secondary structure feature-guided differential evolution named RSDE is proposed to predict three-dimensional structure with the exploitation on the loop region structure. In RSDE, the structure of the loop region is improved by the following: loop-based cross operator, which interchanges configuration of a randomly selected loop region between individuals, and loop-based mutate operator, which considers torsion angle feature into conformational sampling. A stochastic ranking selective strategy is designed to select conformations with low energy and near-native structure. Moreover, the conformational resampling method, which uses previously learned knowledge to guide subsequent sampling, is proposed to improve the sampling efficiency. Experiments on a total of 28 test proteins reveals that the proposed RSDE is effective and can obtain native-like models.

Conformational Space Sampling Method Using Multi-Subpopulation Differential Evolution for De novo Protein Structure Prediction.

Protein structure prediction can be considered as a multimodal optimization problem for sampling the protein conformational space associated with an extremely complex energy landscape. To address this problem, a conformational space sampling method using multi-subpopulation differential evolution, MDE, is proposed. MDE first devotes to generate given numbers of concerned modal under the ultrafast shape recognition-based modal identification protocol, which regards each individual as one modal at beginning. Then, differential evolution is used for keeping the preserved modal survival in the evolution process. Meanwhile, a local descent direction used to sample along with is constructed based on the abstract convex underestimate technique for modal enhancement, which could enhance the ability of sampling in the region with lower energy. Through the sampling process of evolution, several certain clusters contain a series of conformations in proportion to the energy score will be obtained. Representative conformations in the generated clusters can be directly picked out as decoy conformations for further refinement with no extra clustering operation needs. A total of 20 target proteins are tested, in which ten target proteins are tested for comparison with Rosetta and three evolutionary algorithms, and ten easy/hard target proteins in CASP 11 are tested for further verifying the effectiveness of MDE. Test results show strong sampling ability that MDE holds, and near-native conformations can be effectively obtained.

Enhancing Protein Conformational Space Sampling Using Distance Profile-Guided Differential Evolution.

De novo protein structure prediction aims to search for low-energy conformations as it follows the thermodynamics hypothesis that places native conformations at the global minimum of the protein energy surface. However, the native conformation is not necessarily located in the lowest-energy regions owing to the inaccuracies of the energy model. This study presents a differential evolution algorithm using distance profile-based selection strategy to sample conformations with reasonable structure effectively. In the proposed algorithm, besides energy, the residue-residue distance is considered another measure of the conformation. The average distance errors of decoys between the distance of each residue pair and the corresponding distance in the distance profiles are first calculated when the trial conformation yields a larger energy value than that of the target. Then, the distance acceptance probability of the trial conformation is designed based on distance profiles if the trial conformation obtains a lower average distance error compared with that of the target conformation. The trial conformation is accepted to the next generation in accordance with its distance acceptance probability. By using the dual constraints of energy and distance in guiding sampling, the algorithm can sample conformations with lower energies and more reasonable structures. Experimental results of 28 benchmark proteins show that the proposed algorithm can effectively predict near-native protein structures.

A Novel Method Using Abstract Convex Underestimation in Ab-Initio Protein Structure Prediction for Guiding Search in Conformational Feature Space.

To address the searching problem of protein conformational space in ab-initio protein structure prediction, a novel method using abstract convex underestimation (ACUE) based on the framework of evolutionary algorithm was proposed. Computing such conformations, essential to associate structural and functional information with gene sequences, is challenging due to the high-dimensionality and rugged energy surface of the protein conformational space. As a consequence, the dimension of protein conformational space should be reduced to a proper level. In this paper, the high-dimensionality original conformational space was converted into feature space whose dimension is considerably reduced by feature extraction technique. And, the underestimate space could be constructed according to abstract convex theory. Thus, the entropy effect caused by searching in the high-dimensionality conformational space could be avoided through such conversion. The tight lower bound estimate information was obtained to guide the searching direction, and the invalid searching area in which the global optimal solution is not located could be eliminated in advance. Moreover, instead of expensively calculating the energy of conformations in the original conformational space, the estimate value is employed to judge if the conformation is worth exploring to reduce the evaluation time, thereby making computational cost lower and the searching process more efficient. Additionally, fragment assembly and the Monte Carlo method are combined to generate a series of metastable conformations by sampling in the conformational space. The proposed method provides a novel technique to solve the searching problem of protein conformational space. Twenty small-to-medium structurally diverse proteins were tested, and the proposed ACUE method was compared with It Fix, HEA, Rosetta and the developed method LEDE without underestimate information. Test results show that the ACUE method can more rapidly and more efficiently obtain the near-native protein structure.