Current Bioinformatics

With the exploration of big omics data in plants, we are faced with the problem of how to interpret and exploit the rapidly accumulating sets of publicly available omics data. The development and application of computational algorithms, databases and tools are desirable for the efficient processing, management and visualization of large-scale omics data. However, analyzing such big data, deriving biological knowledge, and applying it for predictions and further experimentation back become a challenging tasks. Bioinformatics analyses are important approaches to comprehensively understand how a plant system works by exploiting computational methods to integrate multiple levels of omics datasets. The main objective of this special issue is to provide a forum for researchers to present latest advances and state-of-the-art techniques, tools and applications in analyzing plant omics data. In this issue, a board scope on plant bioinformatics is covered, from genetics, comparative genomics, transcriptomics, gene regulatory networks, signaling pathways, proteomics, non-coding RNAs to high-throughput sequencing analyses and applications. The Single-Molecule Real-Time (SMRT) Isoform Sequencing (Iso-Seq) has paved the way to obtain longer full-length transcripts, with its significance in identifying full-length splice variants and other post-transcriptional events as compared to RNA-Seq. Yubang Gao et al. summarized the existing Iso-Seq analysis tools and presented an integrated bioinformatics pipeline for Iso-Seq analysis, which overcomes the limitations of NGS and generates long contiguous FLNC reads for analysis of post-transcriptional events. Visualization approach of Iso-Seq, and combination of Iso-Seq data with RNA-Seq for transcriptome quantification were also discussed [1]. Pan Wei et al. evaluated the genetic variation is B. germanica based on two mitochondrial genes. Phylogenetic analysis indicated that the B. germanica isolates from central China should be classified as a single population. Demographic analysis rejected the hypothesis of a sudden population expansion of the B. germanica population. These findings indicate that the 36 isolates of B. germanica sampled in the study have high genetic variation, which provides useful knowledge for monitoring changes in parasite populations for future control strategies [2]. Whole genome-wide scans are useful to study positively selected genes (PSGs) to understand the dynamics of genome evolution and the genetic basis of differences between species. Yue Guo et al. investigated the difference of PSGs between two species of cotton. They suggested that the PSGs evolved at a higher rate of synonymous substitutions (Ks), but were subjected to lower selection pressure (Ka/Ks). These findings indicate that PSGs in G. arboreum and G. raimondii differ not only in Ka/Ks, but also in their evolutionary, structural, and expression properties, indicating that the divergence of G. arboreum and G. raimondii were associated with differences in PSGs in terms of evolutionary rates, gene compactness, expression patterns, and WGD retention in Gossypium [3]. Aravind Kumar K et al. identified the differentially expressed WRKY TFs in chickpea in response to herbicide stress and deciphered their interacting partners. Comparison of the differentially expressed TFs, construction of co-functional gene networks, and functional enrichment analysis were conducted. Systems biology approaches reveal a multi-stress responsive WRKY transcription factor and stress associated gene co-expression networks in chickpea [4]. In plant, interactions of various phytohormones play important roles in response to various stresses. Maryam Mortezaeefar et al. investigated the effects of the crosstalk among the hormone signalling pathways in plants. The Weighted Gene Coexpression Network Analysis (WGCNA) method was used to define modules containing genes with highly correlated expression patterns in response to abscisic acid (ABA), jasmonic acid (JA), and salicylic acid (SA) in Arabidopsis. Results indicate that plants create a main change of expression profile and control diverse cell functions, including response to environmental stresses and external factors, cell cycle, and antioxidant activity [5]. Ubiquitination plays a significant role in the regulations of many biological processes, such as cell division, signal transduction, apoptosis and immune response. Identification of ubiquitination sites is the crucial first step to investigate the molecular mechanisms of ubiquitination-related biological processes. Jiajing Chen et al. introduced an ubiquitination site prediction method based on support vector machine (SVM) for Arabidopsis thaliana. The combination of AAC and CKSAAP encoding schemes yielded the best performance with the accuracy and AUC of 81.35% and 0.868 in the independent test [6]. Long noncoding RNAs (lncRNAs), arbitrarily longer than 200 nucleotides, play critical roles in diverse biological processes. Youhuang Bai et al. introduced a database of lncRNA sequences and annotation in plants. PlncRNADB provides a pipeline to quickly distinguish potential noncoding RNAs from protein coding transcripts. The database provides the relationship between lncRNAs and various RNA-binding proteins (RBPs), which can be displayed through a user-friendly web interface. PlncRNADB can serve as a reference database to investigate the lncRNAs and their interaction with RNA-binding proteins in plants [7]. In plant phenomics, one of the key technologies is phenotype image analysis. Bizhi Wu et al. proposed a deep convolutional autoencoder architecture to segment the biological phenotypic images and developed a phenotype retrieval system to enable a better understanding of genotype–phenotype correlation. Their work focused on the identification of similar query phenotypic images by searching the biological phenotype database, including information about loss-of-function and gain-of-function [8]. We hope that the readership of Current Bioinformatics would find this collection of papers interesting and useful.

The increasing role of computer modeling, bioinformatics and operation research in medicine and biology has been remarkable over the last decades or so, owing to the rapid increase in computer processing speed and a greater demand of realistic models capable of predicting actual biological or medical phenomena and optimizing discovery through big data processing. Clearly, naturally occurring biological or physiological processes are complicated and often impossible to be sufficiently simulated or automated ex vivo using 1D or 2D idealized models, let alone its performance optimization. The trend is now moving towards using sophisticated patient-specific real time 3D models; some may even incorporate complex biomaterials or design structures in an effort to produce accurate surrogates to represent biological or medical processes in the laboratory. This drives a greater interest in the optimization of biomaterials, design structures, or biometric techniques with an ultimate aim of repairing or replacement of malfunctioned anatomical structures or translating laboratory research into real life applications. For most of such given problems, multi-objective optimization and data mining can provide the optimal solution using an integrated informatics platform. Moreover, machine learning such as artificial neural network, deep learning, evolutionary algorithm, and genetic algorithm are some of the other well-established computing techniques we can explore for solutions generating solutions. In the medical industry, using these algorithms can help biomedical engineers find the potential factors which affect the process design more accurately and improve bio-device performance. It can also identify trends that bridge the gaps among the fragments of seemingly unrelated information. In addition, the process of information management also promotes the development of bioinformatics, including membrane computing, gene expressions, genetic computing, etc for the development of new products in the medical industry. These new technologies can offer much higher quality and personalized services to humans. This collection of papers in this issue is dedicated to report cutting edge theoretical, experimental, biological, or clinical investigations using advanced biomaterials, design structures, bioinformatics platform, or biometric techniques. We aim at elucidating the complex designs of life through laboratory investigations or computational modeling. We invite submissions from all fields in biomedical engineering, applied biology or medical sciences with a focus on the following: 1) Optimization of flow or structural motions that occur in nature for biometrics and/or bioinformatics studies. 2) Development or optimization of biomaterials that mimic physiological structures, which ultimately aim at accurate understanding of disease processes and in vivo dynamic interactions using a big data processing platform. 3) Novel design or experimental techniques based on multiobjective optimization which allow better understanding of the physiology processes in humans and animals. 4) Understanding physiological processes and incorporating optimization algorithms into the design of devices for superior performance. 5) Evolutionary algorithm or artificial neural network models that help predict disease initiation, progression or end-point to guide and optimize management through informatics. 6) Translation of advanced computational bioinformatics methods into real life applications. The first paper “Bioinformatics study on serum triglyceride levels for analysis of a potential risk factor affecting blood pressure variability” provided by Lin Xu et al. [1] establishes whether triglycerides (TGs) are related to blood pressure (BP) variability and whether controlling TG levels leads to better BP variability management and prevents cardiovascular disease (CVD). In this study, the authors enrolled 106 hypertensive patients and 80 non-hypertensive patients. Pearson correlation and partial correlation analyses were used to define the relationships between TG levels and BP variability in all the subjects. Patients with hypertension were divided into two subgroups according to TG level: Group A (TG<1.7 mmol/L) and Group B (TG>=1.7 mmol/L). The heterogeneity between the two subgroups was compared using t tests and covariance analysis. The second paper was provided by Gang Xu et al. entitled “Bioinformatics study of RNA Interference on the Effect of HIF- 1α on Apelin Expression in Nasopharyngeal Carcinoma Cells” [2]. This paper investigates apelin expression in nasopharyngeal carcinoma CNE-2 cells and its regulation by hypoxia inducible factor-1α (HIF-1α) under hypoxic conditions. CoCl2 was used to induce hypoxia in CNE-2 cells for 12h, 24h and 48h. HIF-1α small interference RNA (siRNA) was transfected into CNE-2 cells using a transient transfection method. HIF-1α and apelin mRNA levels were detected by real time PCR. Western blot was used to measure HIF-1α protein expression. The concentration of apelin in cell culture supernatant was determined by the enzyme linked immunosorbent assay (ELISA). HIF-1α and apelin mRNA levels and protein expression in CNE-2 cells increased gradually with an increased duration of hypoxic exposure and significantly reduced in HIF-1α siRNA transfected cells exposed to the same hypoxic conditions. Apelin expression was induced by hypoxia and regulated by HIF-1α in CNE-2 cells. The third paper provided by Xi-Wen Jiang et al. is entitled as “MGB Blocker ARMS real-time PCR for diagnosis of CYP2C19 mutation in Chinese population” [3]. CYP2C19 is an important genetic factor modulating clopidogrel dose requirement. Therefore, a simple and economic genotyping method for predicting clopidogrel dose of patients would be useful in clinical applications. In this study, the MGB blocker ARMS real-time PCR contained two forward primers and two MGB blockers and a common reverse primer was used for CYP2C19*2, *3 and *17 substitutions. Results showed that heterozygotes and homozygotes of CYP2C19*2, *3 and *17 could be distinguished by the MGB blocker ARMS real-time PCR successfully. In the Chinese population, patients had allele frequencies of CYP2C19*2, *3, and *17 being 18.43%, 3.03% and 0.76%, respectively. This study indicates the MGB blocker ARMS real-time PCR will be a simple, economical method for the rapid detection of SNPs in CYP2C19. In the last paper “Bioinformatics analysis of quantitative PCR and reverse transcription PCR in detecting HCV RNA”, Wei Lie et al. [4] aimed to make comparisons of sensitivity and specificity between quantitative real time polymerase chain reaction (Q-PCR) and reverse transcription PCR (RT-PCR) in detecting the ribonucleic acid (RNA) expression levels of hepatitis C virus (HCV). Patients suffering from hepatitis C and 98 healthy participants with normal liver functions were identified. The venous blood collections were carried out, and were subjected to detect the expression levels of HCV RNA via Q-PCR and RTPCR. After this, the data obtained from the above two detection methods were compared, including the sensitivity and specificity. In summary, these 5 interesting papers provide readers with valuable information on the recent progress in bioinformatics. Hopefully this collection of papers in this issue will provide a platform for researchers, academics and healthcare professionals with the aim of promoting scientific study of bioinformatics. Also, it can be used as a valuable source of references for researchers to produce more promising studies in this field.

In recent years, the use of computational tools and models in conjunction with medical imaging techniques has been effectively employed for analyzing various health conditions. The entropy is a mathematical tool for quantifying the disorder or randomness associated with a system. The entropy can also be interpreted as the degree of uncertainty and hence is a measure of the information content of a signal [1, 2]. The concept of entropy is derived from the second law of thermodynamics and its applications have branched into several areas of science and technology such as signal and image processing and mathematical modelling [3-5]. Recently, various entropy measures such as Tsallis entropy, Renyi entropy, etc. have been successfully utilized for the development of medical diagnostic systems and decision support systems [6, 7]. In the field of image analysis, entropy can provide a good level of information to describe a given image [8]. Entropy measures are highly useful for segmentation, feature extraction and texture characterization of images [9]. Entropy measures have proved to be highly useful for the segmentation of medical images and extraction of valuable features from biomedical signals and images for the development of systems for classification of normal and abnormal cases and also for the analysis of severity of diseases [10, 11]. The Current Bioinformatics Journal offers an excellent forum for dissemination and understanding of interdisciplinary knowledge between the biomedical researchers and the healthcare industry. This special issue of current Bioinformatics entitled “Entropy measures for medical image analysis” unites the authors’ effort to bring recent technological discussion on the applications of entropy measures in the field of medical image processing. This issue comprises three excellent contributions from varied geographical locations around the globe and presents the applications of entropy in diverse medical images such as Microscopic Images, echocardiography ultrasound images and Magnetic Resonance Images (MRI), obtained from various modalities. Liver fibrosis is the formation of scar tissue in dysfunctional liver and is a major health problem in recent years. One of the major causes of liver fibrosis is alcoholism, however several factors such as infections, hereditary causes etc. may also lead to fibrosis of liver. The first paper of this special issue, authored by Yu Wang et al. employs entropy-based features extracted from microscopic images of mice liver for analysis of fibrosis of liver. The second paper in this special issue, authored by Luminita Moraru et al. addresses another important problem dealing with the age-related heart muscle damage. The authors have utilized fuzzy c-means classification algorithm along with texturebased image features such as entropy and homogeneity for classification of healthy subjects and patients with myocardial infarction. In the final paper of this special issue, the authors Suresh Chandra Satapathy et al. have presented a semi-automated examination procedure for inspecting the severity of stroke lesion using MRI images. Also, the authors have employed a novel and new heuristic optimization algorithm namely Social Group Optimization (SGO) algorithm and a thresholding procedure based on Shannon’s entropy, Kapur’s entropy and Otsu’s function for preprocessing the adopted images. This special issue clearly magnifies the applicability and significance of entropy measures for the analysis of medical images. We are indebted to the Editor-in-Chief, Prof. Dr. Yi-Ping Phoebe Chen for her valuable support. We thank Ms. Nida Hatif, Assistant Manager Publications, Bentham Science Publishers for her assistance and support. The editors would like to thank the reviewers of this special issue, for their excellent reviews and comments.

“Prediction”, which is the main task in machine learning research, has now become popular in biomedicine and bioinformatics recently, since Obama proposed the precision medicine project. Precision medicine would combine biomedicine and bioinformatics together, and employ advanced machine learning techniques, especially big data analysis approaches. Traditional medical big data would be calculated with different statistics testing methods, which are difficult to deal with multimodal complex data and the features relationship. Machine learning was considered as a “black box” in the past years. However, it works better than the simple statistic testing methods. In recent years, deep learning and big data machine learning techniques developed fast and showed amazing performance in internet application, understanding images, and auto speech recognition. We can conclude that precision medicine would be the next novel application for the big data machine learning. However, we found that traditional classifiers, such as SVM, random forest, still dominate in the recent biomedicine and bioinformatics works. Therefore, we plan to organize the special issue to bridge the latest methods and the biological applications. In this thematic issue, eight outstanding reviews have been presented from different countries and regions, including P.R. China, USA, and Germany. Jing et al. conducted a systematic review of protein inter-residue contacts prediction methods. As a low-dimensional representation of protein tertiary structure, protein inter-residue contacts could help de novo protein structure prediction methods to reduce the conformational search space. Various methods have been developed for prediction of protein interresidue contacts over the past two decades, and those methods could be roughly classified into five categories: correlated mutations methods, machine-learning methods, fusion methods, template-based methods and 3D model-based methods. In this article, they reviewed three main methods for prediction of protein inter-residue contacts and analyzed constraints of each category. They also compared several representative methods and discussed performances of these methods in detail [1]. Zhang et al. performed a comprehensive review of the recent developments of feature extraction methods for protein sequences, including composition-based features, autocorrelation-based features, and profile-based features. Their concepts and the main equations were introduced and discussed. Finally, some existing computational tools of generating these features were introduced. This article is especially useful for the researchers who are interested in computationally studying the structures and functions of proteins [2]. Liu et al. analyzed the role of bioinformatics in modernization of Traditional Chinese Medicine. Traditional Chinese Medicine (TCM) has a wide range of medicine practices, which was developed in China and other East Asian countries for over 2000 years. The aim of modernization of TCM is to utilize modern biological and medicinal knowledge to research the treatment mechanism of herbal medicine. Combined with rich practice experience of TCM, it can efficiently accumulate development of modern drug. During this process, the bioinformatics as a rising subject can aid to discover and analyze TCM. Currently, bioinformatics has played an important role to understand biological data. In this article, they try to introduce and discuss the applications of bioinformatics techniques in TCM for interested readers and look ahead the potential development of bioinformatics in TCM [3]. Lei et al. conducted a survey on the prediction of essential proteins and genes, which discusses not only the current developments of diverse computational methods, but also the challenge for future work. It is well known that the prediction of essential proteins using computational methods is quite significant in the field of bioinformatics, such as the disease research and drug design. The authors introduced the recent studies for predicting essential proteins and genes by analyzing different kinds of computational methods including topology-based methods, integrated methods, machine learning methods and reliable network methods. In addition, they showed the evaluation techniques of predictive performance and some available databases of essential genes, as well as the future research directions. This review presents the systematic knowledge of the essential proteins and genes prediction which can help the studies in the related fields [4]. Jia and Gong summarized the computational methods in linear B-cell epitope prediction. The determination of B-cell epitopes is thus important for the synthesis of peptide vaccines in the preparation of diagnostic reagents, and the screening of monoclonal antibodies. For identifying such epitopes, many computational predictors have been proposed in the past 10 years. In this review, we summarized the representative computational approaches developed for the identification of linear B-cell epitopes. They mainly discussed the datasets, feature extraction methods and classification methods used in the previous work. Moreover, they also consider existing challenges and future perspectives for developing reliable methods for predicting linear B-cell epitopes [5]. Yang et al. summarized the application of machine learning methods in protein sub-Golgi localization. The distribution of proteins in cell correlates with their functions. Due to its important roles in protein storage, package and distribution, the identification of protein sub-Golgi localization is very important for the in-depth understanding of proteins function. Because transitional biochemical experiments are time-consuming and require expensive materials, machine learning method is used to identify the protein location in Golgi apparatus. In the review, Yang et al. briefly introduced the recent progresses of protein sub-Golgi apparatus localization prediction from the database construction, feature extraction and optimization, prediction algorithms and webserver establishment. The advantages and disadvantages of these methods were discussed. They also provided the prospect of protein sub-Golgi localization prediction using machine learning methods [6]. Zhang et al. analyzed the cross-kingdom regulation of exogenous plant miRNAs in humans. As a new signaling molecule, exogenous miRNAs regulate and influence the physiological functions of inhalers, revealing the effects on human health and metabolism at the gene molecular level. During the process, the combination of bioinformatics and computer methods can help find and analyze the cross-kingdom regulation of exogenous plant miRNA in the human body. In this article, they made attempt to introduce and discuss recent advances of exogenous plant miRNAs into humans and their cross-kingdom regulation and look ahead the potential development of cross-kingdom regulation of exogenous plant miRNAs [7]. Qu et al. analyzed the methods of prediction the DNA-binding protein. DNA-binding protein is an important component of prokaryotic and eukaryotic proteomes, which can participate in many life activities. Due to the importance of DNA-binding protein, there are many researches on prediction DNA-binding, and many methods are mentioned. In the earlier years, the biochemical methods are often used for identification of these proteins, but with the continuous development of machine learning, many machine-learning methods are used to predict DNA-binding proteins. In this article, they analyzed the feature representation methods and common classifiers, summarized the evaluation methods and existing websites, and compared the results of different methods [8]. Each paper in this special issue was extensively peer-reviewed by more than two external reviewers. I would like to thank all the authors for contributing their work to our hot thematic issue and all the reviewers for their time and efforts.

The success of Bioinformatics in recent years has been prompted by research in Molecular Biology and Molecular Medicine in several initiatives. These initiatives gave rise to an exponential increase in the volume and diversification of data, including next generation sequencing data and their annotations, high-throughput experimental (omics) data, biomedical literature, among many others. Systems Biology is a related research area that has been replacing the reductionist” view that dominated Biology research in the last decades, requiring the coordinated efforts of biological researchers with those related to data analysis, mathematical modeling, computer simulation and optimization. The accumulation and exploitation of large-scale databases prompt the development of new computational technology and research on these issues. In this context, many widely successful computational models and tools used by biologists in these initiatives, such as clustering and classification methods for omics data, are based on Computer Science/ Artificial Intelligence (CS/AI) techniques. In fact, these methods have been helping in tasks related to knowledge discovery, modeling and optimization tasks, aiming at the development of computational models so that the response of biological complex systems to any perturbation can be predicted. In this context, the interaction of researchers from different scientific fields is, more than ever, of foremost importance, boosting the research efforts in the field and contributing to the education of a new generation of bioinformatics scientists. The Practical Applications in Computational Biology and Bioinformatics (PACBB) conference has been contributing to this effort, promoting this fruitful interaction over the last 7 years. This special issue gathers four contributions, selected and significantly extended from the rich PACBB'15 technical program, which included papers spanning many different sub-fields in bioinformatics and computational biology. This volume gathers four extended articles selected from the work presented at the PACBB’2015 conference, showing distinct and meaningful practical applications of bioinformatics and computational biology. These range from text mining, to next generation sequencing and gene expression data applications. Calderon-Mantilla et al. propose a pipeline architecture for inferring and visualizing gene networks from expression data, applied to the specific case of coffee plants [1]. Rodriguez-Gonzalez et al. present a comparison of two distinct text mining approaches for extracting diagnostic related knowledge from MedLine Plus articles [2]. The work by Graña et al. proposes nextpresso, a pipeline for the analysis of next generation sequencing (RNA-seq) data that covers the most common requirements of these experimental data [3]. Finally, the work by Fernandez-Gonzalez et al. addresses a relevant problem in text retrieval, namely the influence of class imbalance when developing classifier models for assessing relevant biomedical literature [4].

The major feature of our current life sciences, is the rapid increase of biological data, which are presented in many forms, and reflect the characteristics of biological systems at various levels, including genome, transcriptome, epigenome, proteome and metabolome etc. This is the so-called biological big data we are facing. Biological big data bring both challenges and opportunities to bioinformatics. Tools and techniques for analyzing big biological data enable us to translate massive amount of information into a better understanding of the basic biomedical mechanisms, which can further be applied to translational or personalized medicine. This thematic issue with a theme of “Bioinformatics in Biological Big Data Era” aims at extensively showing the latest development and achievements in Bioinformatics in this biological big data era. The ten papers in this thematic issue were selected from the 1st CCF Bioinformatics Conference (CBC 2016), which was sponsored by China Computer Federation (CCF). The selected papers cover methods and algorithms for processing biological big data in addressing various bioinformatics issues. Before submitted to the special issue, these papers have gone through strict reviewing organized by CBC 2016. Further reviewing was organized by the guest editors. In what follows, we give a brief review of the 10 papers included in this thematic issue. Liu et al. in their paper “Sparse linear modeling kinase inhibition network for predicting combinatorial drug sensitivity in cancer cells” used a sparse linear model called uncertain group sparse representation (UGSR) to infer essential kinases governing the cellular responses to drug treatments, based on the massively collected drug-kinase interactions and drug sensitivity datasets over hundreds of cancer cell lines [1]. In the paper “TagNovo: A dictionary based approach to predict peptide theroy spectra” Wang et al. presented a new theoretical spectrum prediction model called TagNovo, which builds a “tag dictionary" from exiting spectrum library and is used for theory spectrum prediction [2]. In the paper “Large-scale Investigation of Long Noncoding RNA Secondary Structures in Human and Mouse” by Guo et al. the authors conducted a large-scale investigation of lncRNA secondary structures especially for hairpin structural motif in human and mouse based on computational prediction using the RNAfold software, and found that the secondary structures of lncRNAs have many characteristics, most of which are similar with those in mRNAs [3]. Nie et al. in their paper “Prediction of protein S-Sulfenylation sites using a deep belief network” developed a computational method DBN-Sulf to effectively predict S-sulfenylation sites by using optimally extracted properties based on Deep Belief Network (DBN) with Restricted Boltzmann Machines (RBMs). DBN-Suf shows significantly better performance than the existing methods [4]. In the paper “Feature identification for phenotypic classification based on genes and gene pairs”, Su, Zhang and Pan proposed a new algorithm called FSGGP to select both feature genes and feature gene pairs on the binary-value gene expression data [5]. Chan et al. in their paper “MyPhi: Efficient Levenshtein Distance Computation on Xeon Phi based Architectures” introduced MyPhi, an ultra-fast implementation of the Myers algorithm on Intel Xeon Phi based architectures for efficiently computing Levenshtein Distance between genome sequences [6]. The paper “A Metric on the Space of Rooted Phylogenetic Trees” by Wang and Guo proposed a new metric on the space of rooted phylogenetic trees, which can be calculated in polynomial time with the size of the compared trees [7]. Liao et al. developed a method to classify Small GTPases and non-small GTPases in their paper “Classification of small GTPases with hybrid protein features and advanced machine learning techniques” [8]. In the paper “Identification of Attention Deficit/Hyperactivity Disorder in Children Using Multiple ERP Features”, Li et al. used non-invasive event-related potential (ERP) features for Attention deficit hyperactivity disorder (ADHD) prediction [9]. The paper “Low Rank Representation and its application in bioinformatics” by You, Cai and Huang reviews the theoretical and numerical models based on low rank representation and their applications in bioinformatics area [10]. This thematic issue is the result of many people’s contribution, support and cooperation. We appreciate the authors for submitting their works to this thematic issue, and we thank the reviewers for their hard work in reviewing the papers. We also thank the Editor in Chief and the staff of Current Bioinformatics for their valuable help to make this issue possible.

With the development of sequencing technologies, a wide variety of biological data including DNA, RNA and protein sequences and gene expression profiles were generated and accumulated. These data are an external manifestation of acting mechanism of the cell. How to discover cellular mechanism through these data is a vast challenge that current scientists are faced with. The computational approaches including bioinformatics and system biology have proved essential to analyze these complicated data, as Markowetz declared that all biology is computational biology [1]. Recently a number of computational techniques and theories such as BLAST [2, 3] machine learning [4-7] and network theory [8-10], have facilitated the discovery of molecular structures and functions. Therefore, this thematic issue is intended to summarize recent progress of these computational techniques and theories in genomics and proteomics. Comparison between molecular sequences is very helpful to explore evolutionary relationship among different tissues, organisms or species and further to functional analysis. It is not an exaggeration to say that comparison between molecules is an important foundation of life exploration. Natural vector is a method of characterizing protein or DNA sequences and is applicable to classification and evolutionary analysis [11, 12]. Yu [13] reviewed the natural vectors method and application of it in the virus phylogenetic Classification. Biological images are useful especially to phenotype quantification. High-throughput and quantitative biological phenotypes from images are increasingly becoming important to both the quantification of phenotypes and the visualization of biological molecular structure and activity. Bioimage informatics is becoming a new area of exploring life [14]. Chen et al. [15] discussed the major studies based on biological images and summarized the computational techniques of biological image analysis. Long noncoding RNAs (lncRNAs) are transcripts with more than 200 nucleotides, and belong to a type of non-protein coding RNA. LncRNAs have recently been discovered to perform a variety of functions [16]. However, the identification of lncRNA is challenging [17]. Yao et al. [18] reviewed the computational strategies of recognizing lncRNA and current progresses, and discussed existing difficulties in the prediction of the lncRNAs especially by using machine learning methods. Biomedical data are commonly big data which require both high-performance computers and high-effective computational methods. Deep learning proposed by Hinton et al. [19] is becoming a dazzling research field and makes machine intelligence advance a big step. Peng et al. [20] reviewed the application of deep learning in the omics data processing, biological image processing and biomedical diagnosis and discussed challenges. With the development of meteorological science, a large number of meteorology data, such as temperature, humidity, rainfall, air pressure, wind speed and so on, have been collected. So how to identify the key meteorological indexes causing an epidemic? How to integrate these key meteorological indexes to predict the epidemic? These become a vital task for us now. Liu et al. [21] reviewed two categories of model of prediction of the epidemics related to meteorological factors: deterministic models and stochastic models. Single-nucleotide polymorphism known as SNP is referred to as the variation of a single nucleotide that occurs at a specific position in the genome. SNP was found to be associated with a wide range of disease or traits [22], such as inflammatory and autoimmune disorders [23], Alzheimer's disease [24] and breast cancer [25]. The study of SNP-disease associations will facilitate the promise of precision medicine [26]. Li [27] reviewed computational methods of identifying SNP-disease association and discussed improvement directions: data quality improvement, high-performance computing platform and advanced computational method. Although most genes have been detected, little was known about its functions. Numerous computational methods have recently been proposed to find gene functions. Loh et al. [28] summarized these computational methods, compared them and analyzed their strength and weakness. Protein post-translational modification (PTM) is a biochemical reaction which occurs after translation and before protein synthesis, covalently modified by different functional groups. The PTMs involved in every cellular process of life is a key regulating mechanism in the cell [29-32]. The first important step to explore PTMs is to identify PTM types and sites. Currently, the biochemical or biophysical experiments and the computational approaches are parallel to complement one another for identifying PTMs. The computational predictionsgenerally consisted of data collection, representation of PTMs (feature extraction), training the known samples and prediction new samples. Therefore, the feature extraction occupies a central position in the computational prediction of PTMs. Huang [33] reviewed the methods of feature extraction which have recently been developed for PTMs prediction and discussed several properties of it.

No Text Found

No Text Found