DNA methylation is an important biochemical process, and it has a close connection with many types of cancer. Research about DNA methylation can help us to understand the regulation mechanism and epigenetic reprogramming. Therefore, it becomes very important to recognize the methylation sites in the DNA sequence. In the past several decades, many computational methods—especially machine learning methods—have been developed since the high-throughout sequencing technology became widely used in research and industry. In order to accurately identify whether or not a nucleotide residue is methylated under the specific DNA sequence context, we propose a novel method that overcomes the shortcomings of previous methods for predicting methylation sites. We use k-gram, multivariate mutual information, discrete wavelet transform, and pseudo amino acid composition to extract features, and train a sparse Bayesian learning model to do DNA methylation prediction. Five criteria—area under the receiver operating characteristic curve (AUC), Matthew’s correlation coefficient (MCC), accuracy (ACC), sensitivity (SN), and specificity—are used to evaluate the prediction results of our method. On the benchmark dataset, we could reach $0.8632$ on AUC, $0.8017$ on ACC, $0.5558$ on MCC, and $0.7268$ on SN. Additionally, the best results on two scBS-seq profiled mouse embryonic stem cells datasets were $0.8896$ and $0.9511$ by AUC, respectively. When compared with other outstanding methods, our method surpassed them on the accuracy of prediction. The improvement of AUC by our method compared to other methods was at least $0.0399$ . For the convenience of other researchers, our code has been uploaded to a file hosting service, and can be downloaded from: https://figshare.com/s/0697b692d802861282d3.

Background: Protein subcellular localization prediction plays an important role in biology research. Since traditional methods are laborious and time-consuming, many machine learning-based prediction methods have been proposed. However, most of the proposed methods ignore the evolution information of proteins. In order to improve the prediction accuracy, we present a deep learning-based method to predict protein subcellular locations.Results: Our method utilizes not only amino acid compositions sequence but also evolution matrices of proteins. Our method uses a bidirectional long short-term memory network that processes the entire protein sequence and a convolutional neural network that extracts features from protein sequences. The position specific scoring matrix is used as a supplement to protein sequences. Our method was trained and tested on two benchmark datasets. The experiment results show that our method yields accurate results on the two datasets with an average precision of 0.7901, ranking loss of 0.0758 and coverage of 1.2848.Conclusion: The experiment results show that our method outperforms five methods currently available. According to those experiments, we can see that our method is an acceptable alternative to predict protein subcellular location.

Transcription factors (TFs) binding to specific DNA sequences or motifs, are elementary to the regulation of transcription. The gene is regulated by a combination of TFs in close proximity. Analysis of co-TFs is an important problem in understanding the mechanism of transcriptional regulation. Recently, ChIP-seq in mapping TF provides a large amount of experimental data to analyze co-TFs. Several studies show that if two TFs are co-associated, the relative distance between TFs exhibits a peak-like distribution. In order to analyze co-TFs, we develop a novel method to evaluate the associated situation between TFs. We design an adjacency score based on ordered differences, which can illustrate co-TF binding affinities for motif analysis. For all candidate motifs, we calculate corresponding adjacency scores, and then list descending-order motifs. From these lists, we can find co-TFs for candidate motifs. On ChIP-seq datasets, our method obtains best AUC results on five datasets, 0.9432 for NMYC, 0.9109 for KLF4, 0.9006 for ZFX, 0.8892 for ESRRB, 0.8920 for E2F1. Our method has great stability on large sample datasets. AUC results of our method on all datasets are above 0.8.

In the era of big data, the size and complexity of the data are increasing especially for those stored in remote locations, and whose difficulty is further increased by the ongoing rapid accumulation of data scale. Real-world optimization problems present new challenges to traditional intelligent optimization algorithms since the traditional serial optimization algorithm has a high computational cost or even cannot deal with it when faced with large-scale distributed data. Responding to these challenges, a distributed cooperative evolutionary algorithm framework using Spark (SDCEA) is first proposed. The SDCEA can be applied to address the challenge due to insufficient computing resources. Second, a distributed quantum-behaved particle swarm optimization algorithm (SDQPSO) based on the SDCEA is proposed, where the opposition-based learning scheme is incorporated to initialize the population, and a parallel search is conducted on distributed spaces. Finally, the performance of the proposed SDQPSO is tested. In comparison with SPSO, SCLPSO, and SALCPSO, SDQPSO can not only improve the search efficiency but also search for a better optimum with almost the same computational cost for the large-scale distributed optimization problem. In conclusion, the proposed SDQPSO based on the SDCEA framework has high scalability, which can be applied to solve the large-scale optimization problem.

Background: Reconstructing ancestral genomes is one of the central problems presented in genome rearrangement analysis since finding the most likely true ancestor is of significant importance in phylogenetic reconstruction. Large scale genome rearrangements can provide essential insights into evolutionary processes. However, when the genomes are large and distant, classical median solvers have failed to adequately address these challenges due to the exponential increase of the search space. Consequently, solving ancestral genome inference problems constitutes a task of paramount importance that continues to challenge the current methods used in this area, whose difficulty is further increased by the ongoing rapid accumulation of whole-genome data. Results: In response to these challenges, we provide two contributions for ancestral genome inference. First, an improved discrete quantum-behaved particle swarm optimization algorithm (IDQPSO) by averaging two of the fitness values is proposed to address the discrete search space. Second, we incorporate DCJ sorting into the IDQPSO (IDQPSO-Median). In comparison with the other methods, when the genomes are large and distant, IDQPSO-Median has the lowest median score, the highest adjacency accuracy, and the closest distance to the true ancestor. In addition, we have integrated our IDQPSO-Median approach with the GRAPPA framework. Our experiments show that this new phylogenetic method is very accurate and effective by using IDQPSO-Median. Conclusions: Our experimental results demonstrate the advantages of IDQPSO-Median approach over the other methods when the genomes are large and distant. When our experimental results are evaluated in a comprehensive manner, it is clear that the IDQPSO-Median approach we propose achieves better scalability compared to existing algorithms. Moreover, our experimental results by using simulated and real datasets confirm that the IDQPSO-Median, when integrated with the GRAPPA framework, outperforms other heuristics in terms of accuracy, while also continuing to infer phylogenies that were equivalent or close to the true trees within 5 days of computation, which is far beyond the difficulty level that can be handled by GRAPPA.