Search

Jeffrey Gudin,1 Seferina Mavroudi,2,3 Aigli Korfiati,3 Konstantinos Theofilatos,3 Derek Dietze,4 Peter Hurwitz5 1Rutgers New Jersey Medical School, Newark, NJ, USA; 2Department of Nursing, School of Health Rehabilitation Sciences, University of Patras, Pátrai, Greece; 3InSyBio Ltd, Winchester, UK; 4Metrics for Learning LLC, Queen Creek, Arizona, USA; 5Clarity Science LLC, Narragansett, Rhode Island, USACorrespondence: Peter HurwitzClarity Science LLC, 750 Boston Neck Road, Suite 11, Narragansett, RI 02882, USATel +1917 757 0521Fax +1855-891-8303Email peterh@claritysciences.orgPurpose: Chronic pain is a life changing condition, and non-opioid treatments have been lately introduced to overcome the addictive nature of opioid therapies and their side effects. In the present study, we explore the potential of machine learning methods to discriminate chronic pain patients into ones who will benefit from such a treatment and ones who will not, aiming to personalize their treatment.Patients and Methods: In the current study, data from the OPERA study were used, with 631 chronic pain patients answering the Brief Pain Inventory (BPI) validated questionnaire along with supplemental questions before and after a follow-up period. A novel machine learning approach combining multi-objective optimization and support vector regression was used to build prediction models which can predict, using responses in the baseline, the four different outcomes of the study: total drugs change, total interference change, total severity change, and total complaints change. Data were split to training (504 patients) and testing (127 patients) sets and all results are measured on the independent test set.Results: The machine learning models extracted in the present study significantly overcame other state of the art machine learning methods which were deployed for comparative purposes. The experimental results indicated that the machine learning models can predict the outcomes of this study with considerably high accuracy (AUC 73.8– 87.2%) and this allowed their incorporation in a decision support system for the selection of the treatment of chronic pain patients.Conclusion: Results of this study revealed the potential of machine learning for an individualized medicine application for chronic pain therapies. Topical analgesics treatment were provento be, in general, beneficial but carefully selecting with the suggested individualized medicine decision support system was able to decrease by approximately 10% the patients which would have been subscribed with topical analgesics without having benefits from it.Keywords: individualized medicine, pain therapy, non-opioid treatment, machine learning, predictive analytics, regression, multi-objective optimization, support vector regression

ImportancePleural effusions frequently signal disseminated cancer. Diagnostic markers of pleural malignancy at presentation that would assess cancer risk and would streamline diagnostic decisions remain unidentified.ObjectiveThe present study aimed at identifying and validating predictors of malignant pleural effusion at patient presentation.DESIGN, SETTING, AND PARTICIPANTSA consecutive cohort of 323 patients with pleural effusion (PE) from different etiologies were recruited between 2013-2017 and was retrospectively analyzed. Data included history, chest X-ray, and blood/pleural fluid cell counts and biochemistry. Group comparison, receiver-operator characteristics, unsupervised hierarchical clustering, binary logistic regression, and random forests were used to develop the malignant pleural effusion detection (MAPED) score. MAPED was validated in an independent retrospective UK cohort (n= 238).Main Outcomes and MeasuresThe outcome was diagnostic of pleural effusion in patients, and the clinical and laboratory indicators available of the patient were measured.ResultsFive variables showed significant diagnostic power and were incorporated into the 5-point MAPED score. Age > 55 years, effusion size > 50% of the most affected lung field, pleural neutrophil count < 2,500/mm3, effusion protein > 3.5 g/dL, and effusion lactate dehydrogenase > 250 U/L, each scoring one point, predicted underlying cancer with the area under curve(AUC) = 0.819 (sensitivity=82%, specificity=74%,P< 10-15) in the derivation cohort. The AUC and net reclassification improvement (NRI) of MAPED score and cytology were not significantly different. However, the integrated discrimination improvement (IDI) of The MAPED score displayed a slight increment(PConclusionsThe MAPED score identifies malignant pleural effusions with satisfactory accuracy and can be used complementary to cytology to streamline diagnostic procedures.

Background Identifying molecular biomarkers characteristic of ischemic stroke has the potential to aid in distinguishing stroke cases from stroke mimicking symptoms, as well as advancing the understanding of the physiological changes that underlie the body’s response to stroke. This study uses machine learning-based analysis of gene co-expression to identify transcription patterns characteristic of patients with acute ischemic stroke. Methods Mutual information values for the expression levels among 13,243 quantified transcripts were computed for blood samples from 82 stroke patients and 68 controls to construct a co-expression network of genes (separately) for stroke and control samples. Page rank centrality scores were computed for every gene; a gene’s significance in the network was assessed according to the differences in their network’s pagerank centrality between stroke and control expression patterns. A hybrid genetic algorithm – support vector machine learning tool was used to classify samples based on gene centrality in order to identify an optimal set of predictor genes for stroke while minimizing the number of genes in the model. Results A predictive model with 89.6% accuracy was identified using 6 network-central and differentially expressed genes (ID3, MBTPS1, NOG, SFXN2, BMX, SLC22A1), characterized by large differences in association network connectivity between stroke and control samples. In contrast, classification models based solely on individual genes identified by significant fold-changes in expression level provided lower predictive accuracies

Chronic pain has been linked with negative impacts on psychological and social factors, with mortality and several diseases. Lately, emphasis has been focused on non-opioid treatments to overcome its addictive nature and other side effects. To address this, the OPERA study evaluated the effectiveness of topical analgesics as an alternative method to opioids for pain therapy. Initial results showed that topical analgesics have significant benefits for the majority of chronic pain patients. However, there were segments of patients who did not seem to benefit from prescribed therapy and some participants whose situation deteriorated after the intervention. In the present study, we are exploring the potential of machine learning methods to classify chronic pain patients into those who will benefit from topical analgesics treatment and those who will not, in order to personalize their treatment. For this purpose, we utilized a hybrid combination of multi-objective optimization and support vector regression which is able to handle imbalanced datasets, select the optimal features subset and optimize the parameters of the regression model so as to maximize the predictive accuracy. The proposed method significantly overcame other state-of-the-art methods. Experimental results showed that its application can predict, with reasonable accuracy (AUC 73.8-87.2%), the outcomes of this study allowing for a precision medicine approach in treating chronic pain.

Introduction: Identifying the molecular biomarkers characteristic of ischemic stroke has the potential to aid in distinguishing stroke cases from stroke symptom mimics while advancing the understanding of the physiological changes underlying the body's response to stroke. Hypothesis: Measures of network centrality can identify a more statistically robust and functionally significant set of molecular biomarkers to distinguish between stroke and non-stroke cases than methods based solely on the absolute magnitudes of gene expression fold-change. Methods: Mutual information values for the expression levels of 13243 quantified transcripts were evaluated in blood samples from 128 stroke patients and 67 non-stroke patients to construct co-expression networks of genes. Page rank centrality scores were computed for every gene; a gene’s significance in the network was assessed according to the differences in their network centrality between stroke and non-stroke expression patterns. A hybrid genetic algorithm-support vector machine learning tool (InSyBio) was used to classify samples based on gene centrality in order to identify an optimal set of predictor genes for stroke vs. non-stroke while minimizing the number of genes in the model. Results: A predictive model for stroke with 85.7% accuracy was found using 4 network-central genes ( AGTPBP1, SRPK1, ANTXR2, MRPL41 ). In contrast, classification models based solely on individual genes identified by significant fold-change in expression level provided predictive accuracies < 70% for any single gene. Models with much larger (10-26) numbers of gene transcript biomarkers gave lower predictive accuracies (≤ 80%) than the 4 network-based gene classification in this study. The 4 genes identified from network centrality were also significantly associated with differential expression in 6 miRNAs, and were involved in key pathways including metal ion binding, translation, and transport across cell/organelle membranes. Conclusions: This study provides an improved set of biomarkers that may be clinically useful for diagnosing stroke. Additionally, the set of co-expressed genes identified in this study provides insight into novel pathways that are activated or disrupted as a consequence of stroke.

This file summarizes analyses comparing pooled whole blood + PBMC samples vs. whole blood only. Figure S1.1 shows a volcano plot of FC vs. log p-values for the blood samples only, which is qualitatively consistent with Fig. 1 for pooled data. The second figure is a scatterplot and Spearman correlation analysis of pooled data FC vs. blood only FC, confirming consistency across the datasets. (DOCX 36 kb)

MicroRNAs (miRNAs) are small non-coding RNAs, which play a significant role in gene regulation. Predicting miRNA genes is a challenging bioinformatics problem and existing experimental and computational methods fail to deal with it effectively. We developed YamiPred, an embedded classification method that combines the efficiency and robustness of support vector machines (SVM) with genetic algorithms (GA) for feature selection and parameters optimization. YamiPred was tested in a new and realistic human dataset and was compared with state-of-the-art computational intelligence approaches and the prevalent SVM-based tools for miRNA prediction. Experimental results indicate that YamiPred outperforms existing approaches in terms of accuracy and of geometric mean of sensitivity and specificity. The embedded feature selection component selects a compact feature subset that contributes to the performance optimization. Further experimentation with this minimal feature subset has achieved very high classification performance and revealed the minimum number of samples required for developing a robust predictor. YamiPred also confirmed the important role of commonly used features such as entropy and enthalpy, and uncovered the significance of newly introduced features, such as %A-U aggregate nucleotide frequency and positional entropy. The best model trained on human data has successfully predicted pre-miRNAs to other organisms including the category of viruses.

EE-MC is a new unsupervised methodology for predicting protein complexes from weighted PPI graphs.It is by design able to overcome intrinsic limitations of existing methodologies.It outperformed existing methodologies increasing the separation metric by 10-20%.72.58% of the predicted protein complexes in human are enriched for at least one GO function term. ObjectiveProteins are considered to be the most important individual components of biological systems and they combine to form physical protein complexes which are responsible for certain molecular functions. Despite the large availability of protein-protein interaction (PPI) information, not much information is available about protein complexes. Experimental methods are limited in terms of time, efficiency, cost and performance constraints. Existing computational methods have provided encouraging preliminary results, but they phase certain disadvantages as they require parameter tuning, some of them cannot handle weighted PPI data and others do not allow a protein to participate in more than one protein complex. In the present paper, we propose a new fully unsupervised methodology for predicting protein complexes from weighted PPI graphs. Methods and materialsThe proposed methodology is called evolutionary enhanced Markov clustering (EE-MC) and it is a hybrid combination of an adaptive evolutionary algorithm and a state-of-the-art clustering algorithm named enhanced Markov clustering. EE-MC was compared with state-of-the-art methodologies when applied to datasets from the human and the yeast Saccharomyces cerevisiae organisms. ResultsUsing public available datasets, EE-MC outperformed existing methodologies (in some datasets the separation metric was increased by 10-20%). Moreover, when applied to new human datasets its performance was encouraging in the prediction of protein complexes which consist of proteins with high functional similarity. In specific, 5737 protein complexes were predicted and 72.58% of them are enriched for at least one gene ontology (GO) function term. ConclusionsEE-MC is by design able to overcome intrinsic limitations of existing methodologies such as their inability to handle weighted PPI networks, their constraint to assign every protein in exactly one cluster and the difficulties they face concerning the parameter tuning. This fact was experimentally validated and moreover, new potentially true human protein complexes were suggested as candidates for further validation using experimental techniques.

The discovery of miRNAs had a great impact on traditional biology. Typically, miRNAs have the potential to bind to the 3'untranslated region (UTR) of their mRNA target genes for cleavage or translational repression. The experimental identification of their targets has many drawbacks including cost, time and low specificity and these are the reasons why many computational approaches have been developed so far. However, existing computational approaches do not include any advanced feature selection technique and they are facing problems concerning their classification performance and their interpretability. In the present paper, we propose a computational intelligent framework to overcome existing difficulties in predicting miRNA targets. This framework includes three distinct steps: a filtering step, a novel hybrid classification methodology and an advanced methodology to extract interpretable fuzzy rules from the final prediction model. The classification methodology relies on a combination of Genetic Algorithms and Support Vector Machines in order to locate the optimal feature subset while achieving high classification performance. The proposed methodology has been compared with several existing methodologies, a very popular artificial neural network based classification technique and a traditional fuzzy rule based classification method. It was proved that the new method outperforms them in terms of classification performance while it selects a much smaller feature subset and it is able to uncover meaningful biological conclusions.

ScalaLab and GroovyLab are both MATLAB-like environments for the Java Virtual Machine. ScalaLab is based on the Scala programming language and GroovyLab is based on the Groovy programming language. They present similar user interfaces and functionality to the user. They also share the same set of Java scientific libraries and of native code libraries. From the programmer's point of view though, they have significant differences. This paper compares some aspects of the two environments and highlights some of the strengths and weaknesses of Scala versus Groovy for scientific computing. The discussion also examines some aspects of the dilemma of using dynamic typing versus static typing for scientific programming. The performance of the Java platform is continuously improved at a fast pace. Today Java can effectively support demanding high-performance computing and scales well on multicore platforms. Thus, both systems can challenge the performance of the traditional C/C++/Fortran scientific code with an easier to use and more productive programming environment.

Several techniques have been applied on leakage current waveforms in order to extract information regarding electrical activity on high-voltage insulators. However, a fully representative value is yet to be defined. In this article, a hybrid support vector fuzzy inference system is introduced as a classification tool. The system incorporates fuzzy logic, genetic algorithms, and support vector machines. Apart from the classification accuracy achieved, the system also produces a set of fuzzy rules under which the classification is made, allowing a further insight of the process. A comparison is made to other classification tools previously applied on the same data set.

Proteins are the functional components of many cellular processes and the identification of their physical protein---protein interactions (PPIs) is an area of mature academic research. Various databases have been developed containing information about experimentally and computationally detected human PPIs as well as their corresponding annotation data. However, these databases contain many false positive interactions, are partial and only a few of them incorporate data from various sources. To overcome these limitations, we have developed HINT-KB ( http://biotools.ceid.upatras.gr/hint-kb/ ), a knowledge base that integrates data from various sources, provides a user-friendly interface for their retrieval, calculates a set of features of interest and computes a confidence score for every candidate protein interaction. This confidence score is essential for filtering the false positive interactions which are present in existing databases, predicting new protein interactions and measuring the frequency of each true protein interaction. For this reason, a novel machine learning hybrid methodology, called (Evolutionary Kalman Mathematical Modelling--EvoKalMaModel), was used to achieve an accurate and interpretable scoring methodology. The experimental results indicated that the proposed scoring scheme outperforms existing computational methods for the prediction of PPIs.

MicroRNAs transcription regulation is an open topic in molecular biology and the identification of the promoters of microRNAs would give us relevant insights on cellular regulatory mechanisms. In the present study, we introduce a new computational methodology for the prediction of microRNA promoters, which is based on the hybrid combination of an adaptive genetic algorithm with a nu-Support Vector Regression (nu-SVR) classifier. This methodology uses genetic algorithms to locate the optimal features set and to optimize the parameters of the nu-SVR classifier. The main advantage of the proposed solution is that it systematically studies and calculates a vast number of features that can be used for promoters prediction including frequency-based properties, regulatory elements and epigenetic features. The proposed method also handles efficiently the issues of over-fitting, feature selection, convergence and class imbalance. Experimental results give accuracy over 87 % in the miRNA promoter prediction.

The analysis of protein-protein interactions (PPIs) is crucial to the understanding of cellular processes. In recent years, a variety of computational methods have been developed to supplement the interactions that have been detected experimentally. The article’s main objective is to present a novel classification framework for predicting PPIs combining the advantages of two algorithmic methods’ categories. State-of-the-art adaptive filtering techniques were combined with the most contemporary heuristic methods which are based in the natural selection process. The authors’ goal is to find a simple mathematical equation that governs the best classifier enabling the extraction of biological knowledge. The proposed methodology assigns a confidence score to each protein pair and as a result a weighted PPI network is constructed. All possible combinations of the selected adaptive filtering and heuristic techniques were used and comparisons were made to explore the classifiers with the highest performance and interpretability.

By modifying Groovy with Matlab-like constructs, the authors created a compiled mathematical scripting language called GroovySci for the jLab platform. The resulting code generation enhancements could ultimately extend Java's potential for scientific computing.

Gradually Expanding Dense Neighborhoods (GENA) is proposed for the computational prediction of protein complexes from weighted protein interaction networks.GENA permits the participation of proteins to multiple complexes in agreement with the underlying cell mechanisms.GENA outperformed three of the state of the art algorithms for predicting protein complexes in experiments with datasets from yeast and human organisms.Downstream analysis of the resulted clusters revealed functional homogeneity between the proteins of the same cluster.Significantly altered network modules were detected when GENA was applied to two co-expression networks: one generated from Parkinson patients and one from healthy individuals. ObjectiveProteins are vital biological molecules driving many fundamental cellular processes. They rarely act alone, but form interacting groups called protein complexes. The study of protein complexes is a key goal in systems biology. Recently, large protein-protein interaction (PPI) datasets have been published and a plethora of computational methods that provide new ideas for the prediction of protein complexes have been implemented. However, most of the methods suffer from two major limitations: First, they do not account for proteins participating in multiple functions and second, they are unable to handle weighted PPI graphs. Moreover, the problem remains open as existing algorithms and tools are insufficient in terms of predictive metrics. MethodIn the present paper, we propose gradually expanding neighborhoods with adjustment (GENA), a new algorithm that gradually expands neighborhoods in a graph starting from highly informative "seed" nodes. GENA considers proteins as multifunctional molecules allowing them to participate in more than one protein complex. In addition, GENA accepts weighted PPI graphs by using a weighted evaluation function for each cluster. ResultsIn experiments with datasets from Saccharomyces cerevisiae and human, GENA outperformed Markov clustering, restricted neighborhood search and clustering with overlapping neighborhood expansion, three state-of-the-art methods for computationally predicting protein complexes. Seven PPI networks and seven evaluation datasets were used in total. GENA outperformed existing methods in 16 out of 18 experiments achieving an average improvement of 5.5% when the maximum matching ratio metric was used. Our method was able to discover functionally homogeneous protein clusters and uncover important network modules in a Parkinson expression dataset. When used on the human networks, around 47% of the detected clusters were enriched in gene ontology (GO) terms with depth higher than five in the GO hierarchy. ConclusionsIn the present manuscript, we introduce a new method for the computational prediction of protein complexes by making the realistic assumption that proteins participate in multiple protein complexes and cellular functions. Our method can detect accurate and functionally homogeneous clusters.

High throughput sequencing technologies alongside with advanced bioinformatics methods have uncovered a vast number of short non-coding RNAs. They are split in various categories based on their cellular functionality and their sequential, thermodynamic and structural properties. Existing computational methods emphasize on the prediction of only one type of non-coding RNAs and thus their applicability in full transcriptome studies is limited. Only a few methods so far have been proposed for predicting multiple short non-coding RNAs and they do not cover the most significant classes of them. In the present paper, we introduce a new multiclass method based on the combination of genetic algorithms and support vector machines which is able to distinguish among tRNAs, miRNAs, snoRNAs, rRNAs and other RNA sequences with accuracy over 93%. Finally, the advanced feature selection mechanism of the proposed method uncovers significant characteristics for each one of the studied short non-coding RNAs.

Proteins have been proven to be among the most significant cellular molecules as they participate in most cellular functionalities. Researchers have deployed a variety of experimental methods for the identification of Protein–Protein Interactions (PPIs). The emergence of high-throughput experimental techniques for the prediction of PPIs, revealed a wide range of PPIs in many organisms. This information alongside with information from small scale experimental techniques has been stored in public available databases and repositories. It is well-known that experimental data include many false positive predictions and provide only low coverage on the full interactomes. This fact has led to the design and development of many computational methods for the prediction of PPIs (Theofilatos et al., 2011). The experimental PPI data have been extensively used in many studies during the last decades and their availability gave a significant boost in training new algorithmic models for the prediction of PPIs and in the overall analysis of PPI data. Despite the promising results of algorithmic solutions for PPIs' analysis which fostered molecular biology research, in our opinion the research on computational methods for analyzing PPI data has been recently stagnated. Using the online tool MLtrends introduced by Palidwor and Andrade-Navarro (2010) in a preliminary investigation we have observed that the publications related to the search term “protein–protein interactions” AND analysis (abstract and title were searched for this term) present a constant increase in absolute numbers. However, when applying normalization by dividing with the total number of annual publications, we observe a relatively stable percentage of publications related PPI analysis in the last decade. In contrast, systems biology publications present a big positive slope in the last decade even when normalized by the total number of annual publications. This diversification shows that even if the actual total number of publications related to PPIs analysis is increasing as the total number of scientific journals is increasing in the last decade, their total impact on the systems biology domain is decreasing. Additionally, only a few PPI based research works have been published lately with significant impact in clinical research and translational bioinformatics. In this paper, first we summarize the developments on computational analysis of PPI data and second, we present our belief about the future of PPI data analysis emphasizing in presenting the constraints that have delayed the transition from the current methodologies to a holistic bioinformatics approach, for linking biological and clinical data. Specific solutions are also proposed for all these constraints in order to achieve the optimal exploitation of PPI bioinformatics' approaches.

Fuzzy association rules can reveal useful dependencies and interactions hidden in large gene expression data sets. However their derivation perplexes very difficult combinatorial problems that depend heavily on the size of these sets. The paper follows a divide and conquer approach to the problem that obtains computationally manageable solutions. Initially we cluster genes that more probably are associated. Thereafter, the fuzzy association rule extraction algorithms confront many but significantly reduced computationally problems that usually can be processed very fast. The clustering phase is accomplished by means of an approach based on mutual information (MI). This approach uses the mutual information as a similarity measure. However, the numerical evaluation of the MI is subtle. We experiment with the main methods and we compare between them. As the device that implements the mutual information clustering we use a SOM (Self-Organized Map) based approach that is capable of effectively incorporating supervised bias. After the mutual information clustering phase the fuzzy association rules are extracted locally on a per cluster basis. The paper presents an application of the techniques for mining the gene expression data. However, the presented techniques can easily be adapted and can be fruitful for intelligent exploration of any other similar data set as well.

The application of the Radial Basis Function neural networks in domains involving prediction and classification of symbolic data requires a reconsideration and a careful definition of the concept of distance between patterns. This distance in addition to providing information about the proximity of patterns should also obey some mathematical criteria in order to be applicable. Traditional distances are inadequate to access the differences between symbolic patterns. This work proposes the utilization of a statistically extracted distance measure for Generalized Radial Basis Function (GRBF) networks. The main properties of these networks are retained in the new metric space. Especially, their regularization potential can be realized with this type of distance. However, the examples of the training set for applications involving symbolic patterns are not all of the same importance and reliability. Therefore, the construction of effective decision boundaries should consider the numerous exceptions to the general motifs of classification that are frequently encountered in data mining applications. The paper supports that heuristic Instance Based Learning (IBL) training approaches can uncover information within the uneven structure of the training set. This information is exploited for the estimation of an adequate subset of the training patterns serving as RBF centers and for the estimation of effective parameter settings for those centers. The IBL learning steps are applicable to both the traditional and the statistical distance metric spaces and improve significantly the performance in both cases. The obtained results with this two-level learning method are significantly better than the traditional nearest neighbour schemes in many data mining problems.

Complex application domains involve difficult pattern classification problems. The state space of these problems consists of regions that lie near class separation boundaries and require the construction of complex discriminants while for the rest regions the classification task is significantly simpler. The motivation for developing the Supervised Network Self-Organizing Map (SNet-SOM) model is to exploit this fact for designing computationally effective solutions. Specifically, the SNet-SOM utilizes unsupervised learning for classifying at the simple regions and supervised learning for the difficult ones in a two stage learning process. The unsupervised learning approach is based on the Self-Organizing Map (SOM) of Kohonen. The basic SOM is modified with a dynamic node insertion/deletion process controlled with an entropy based criterion that allows an adaptive extension of the SOM. This extension proceeds until the total number of training patterns that are mapped to neurons with high entropy (and therefore with ambiguous classification) reduces to a size manageable numerically with a capable supervised model. The second learning phase (the supervised training) has the objective of constructing better decision boundaries at the ambiguous regions. At this phase, a special supervised network is trained for the computationally reduced task of performing the classification at the ambiguous regions only. The performance of the SNet-SOM has been evaluated on both synthetic data and on an ischemia detection application with data extracted from the European ST-T database. In all cases, the utilization of SNet-SOM with supervised learning based on both Radial Basis Functions and Support Vector Machines has improved the results significantly related to those obtained with the unsupervised SOM and has enhanced the scalability of the supervised learning schemes. The highly disciplined design of the generalization performance of the Support Vector Machine allows to design the proper model for the particular training set.

Non-coding RNA (ncRNA) genes are RNA molecules transcribed from DNA, but not translated to proteins. Instead, they are involved in gene regulatory mechanisms and a variety of other cellular processes, and they have been associated with several diseases. Thus, their accurate and efficient identification as well as the identification of the genes they regulate is a promising research area and an open topic for the bioinformatics community. In the present paper, we present InSyBio ncRNASeq, which is a cloud-based web platform assisting users to analyze ncRNA sequences and predict whether they are miRNAs, pseudo-hairpins or if they belong to another ncRNA category. Additionally, InSyBio ncRNASeq offers a unique miRNA target prediction pipeline, which results in scored miRNA target sites in genes’ 3’UTRs. This tool not only presents the highest accuracy metrics compared with other state of the art methods and tools, but also offers batch searches between many or all known human miRNAs and many mRNAs.

Computational intelligence, a sub-branch of artificial intelligence, is a field which draws on the natural world and adaptive mechanisms in order to study behaviour in changing complex environments. This book provides an interdisciplinary view of current technological advances and challenges concerning the application of computational intelligence techniques to financial time-series forecasting, trading and investment. The book is divided into five parts. The first part introduces the most important computational intelligence and financial trading concepts, while also presenting the most important methodologies from these different domains. The second part is devoted to the application of traditional computational intelligence techniques to the fields of financial forecasting and trading, and the third part explores the applications of artificial neural networks in these domains. The fourth part delves into novel evolutionary-based hybrid methodologies for trading and portfolio management, while the fifth part presents the applications of advanced computational intelligence modelling techniques in financial forecasting and trading. This volume will be useful for graduate and postgraduate students of finance, computational finance, financial engineering and computer science. Practitioners, traders and financial analysts will also benefit from this book.

Proteins and their interactions have been proven to play a central role in many cellular processes and have been extensively studied so far. However of great importance, little work has been conducted for the identification of biological process interactions in the higher cellular level which could provide knowledge about the high level cellular functionalities and maybe enable researchers to explain mechanisms that lead to diseases. Existing computational approaches for predicting Biological Process interactions used PPI graphs of low quality and coverage but failed to utilize weighted PPI graphs to quantify the quality of the interactions. In the present paper, we propose a unified two-step framework to reach the goal of predicting biological process interactions. After conducting a comparative study we selected as a first step the EVOKALMA model as a very promising algorithm for robust PPI prediction and scoring. Then, in order to be able to handle weights, we combined it with a novel variation of an existing algorithm for predicting biological processes interactions. The overall methodology was applied for predicting biological processes interactions for Saccharomyces Cerevisiae and Homo Sapiens organisms, uncovering thousands of interactions for both organisms. Most of the linked processes come in agreement with the existing knowledge but many of them should be further studied.

Although there are a lot of robust and effective scientific libraries in Java, the utilization of these libraries in pure Java is difficult and cumbersome, especially for the average scientist that does not expertise in software development. We illustrate that ScalaLab presents an easier and productive MATLAB like front end. Also, the main strengths and weaknesses of the core Java libraries of ScalaLab are elaborated. Since performance is of paramount importance for scientific computation, the article discusses extensively performance aspects of the ScalaLab environment. Also, Java bytecode performance is compared to native code.

We present a new probabilistic symmetric key encryption scheme based on the chaotic dynamics of properly designed chaotic systems. This technique exploits the concept of virtual attractors, which are not real attractors of the underlying chaotic dynamics but are created and maintained artificially. Each virtual attractor represents a symbol of the alphabet used to encode messages. The state space is partitioned over the virtual attractors creating clusters of states. The enciphering process randomizes over the set of states mapped to a virtual attractor in order to construct the ciphertext for the transmited symbol. The receiver can reconstruct perfectly this virtual state space, given the possession of the same chaotic system of difference equations with parameters tuned perfectly to those of the transmitter. Therefore, from the ciphertext chunk corresponding to a state, the virtual attractor can be derived from the details of the virtual state space. The knowledge of the virtual attractor leads to the recovery of the transmitted symbol. We demonstrate that the new algorithm is secure, reliable and very fast. It uses discrete time chaotic recurrent systems and is simple, flexible and modular. These systems can be constructed easily dynamically from an alphanumeric encryption key. The cryptographic security of the algorithm is evaluated with combinatorial arguments.

Proteins and their interactions have been proven to play a central role in many cellular processes. Although there are many experimental techniques for protein-protein interaction prediction, only a few exist for predicting protein complexes. For the sake of this, researchers have emphasized lately in the computational prediction of protein complexes from Protein-Protein Interaction (PPI) data. The two major limitations of the current advances in the prediction of protein complexes are that most of the algorithms do not take into consideration the participation of a protein to many protein complexes and that they cannot handle weighted PPI graphs. In the present paper, we altered the original Restricted Neighborhood Search Clustering (RNSC) algorithm to overcome the above limitations. The Enhanced Weighted Restricted Neighborhood Search Clustering (EWRNSC) permits the participation of a protein to many protein complexes by modifying the moves of the original RNSC. In addition, EWRNSC can accept and process weighted PPI graphs as inputs by altering the cost functions of the original RNSC cost clustering schemes. When experimented using atasets from Human, the proposed algorithm proved to outperform the original RNSC and the MCL algorithms which are two of the most broadly used methods in the field of protein complexes prediction.

Biological networks have been widely used in systems biology in order to model the complex interactions of molecular players such as proteins, genes, mRNAs, non-coding RNAs and others. However, most of the current methods for biomarker discovery do not use biological networks, but just deploy simple statistical methods to identify differentially expressed genes and gene products. In the present paper, we present InSyBio BioNets, which is a cloud-based web platform offering a unique biomarker discovery pipeline, which combines differential expression analysis and a method for comparing biological networks to identify biomarkers efficiently. As a case study, InSyBio BioNets was applied to a Parkinson dataset of gene expression measurements and outperformed a standard statistical approach by recovering a more compact and informative set of biomarkers.

Traditional biology was forced to restate some of its principles when the microRNA (miRNA) genes and their regulatory role were firstly discovered. Typically, miRNAs are small non-coding RNA molecules which have the ability to bind to the 3'untraslated region (UTR) of their mRNA target genes for cleavage or translational repression. Existing experimental techniques for their identification and the prediction of the target genes share some important limitations such as low coverage, time consuming experiments and high cost reagents. Hence, many computational methods have been proposed for these tasks to overcome these limitations. Recently, many researchers emphasized on the development of computational approaches to predict the participation of miRNA genes in regulatory networks and to analyze their transcription mechanisms. All these approaches have certain advantages and disadvantages which are going to be described in the present survey. Our work is differentiated from existing review papers by updating the methodologies list and emphasizing on the computational issues that arise from the miRNA data analysis. Furthermore, in the present survey, the various miRNA data analysis steps are treated as an integrated procedure whose aims and scope is to uncover the regulatory role and mechanisms of the miRNA genes. This integrated view of the miRNA data analysis steps may be extremely useful for all researchers even if they work on just a single step.

Proteins and their interactions have been proven to play a central role in many cellular processes. Thus, many experimental methods have been developed for their prediction. These experimental methods are uneconomic and time consuming in the case of low throughput methods or inaccurate in the case of high throughput methods. To overcome these limitations, many computational methods have been developed to predict and score Protein-Protein Interactions (PPIs) using a variety of functional, sequential and structural data for each protein pair. Existing computational methods can still be enhanced in terms of classification performance and interpretability. In the present paper we present a novel Gene Expression Programming (GEP) algorithm, named as jGEPModelling 2.0, and apply it to the problem of PPI prediction and scoring. jGEPModelling2.0 is a variation of the classic GEP algorithm to make it suitable for the problem of PPI prediction and enhance its classification performance. To test its efficiency, we applied it to a public available dataset and compared it to two other state-of-the-art PPI prediction models. Experimental results proved that jGEPModelling2.0 outperformed existing methodologies in terms of classification performance and interpretability. (This paper is submitted for the CIAB2012 workshop).

Until recently, it was commonly accepted that most genetic information is transacted by proteins. Recent evidence suggests that the majority of the genomes of mammals and other complex organisms are in fact transcribed into non-coding RNAs (ncRNAs), many of which are alternatively spliced and/or processed into smaller products. Non coding RNA genes analysis requires the calculation of several sequential, thermodynamical and structural features. Many independent tools have already been developed for the efficient calculation of such features but to the best of our knowledge there does not exist any integrative approach for this task. The most significant amount of existing work is related to the miRNA class of non-coding RNAs. MicroRNAs (miRNAs) are small non-coding RNAs that play a significant role in gene regulation and their prediction is a challenging bioinformatics problem. Non-coding RNA feature extraction and pre-miRNA classification Web Tool (ncRNA-class Web Tool) is a publicly available web tool ( http://150.140.142.24:82/Default.aspx ) which provides a user friendly and efficient environment for the effective calculation of a set of 58 sequential, thermodynamical and structural features of non-coding RNAs, plus a tool for the accurate prediction of miRNAs.

Proteins and their interactions are considered to play a significant role in many cellular processes. The identification of Protein-Protein interactions (PPIs) in human is an open research area. Many Databases, which contain information about experimentally and computationally detected human PPIs as well as their corresponding annotation data, have been developed. However, these databases contain many false positive interactions, are partial and only a few of them incorporate data from various sources. To overcome these limitations, we have developed HINT-KB (http://150.140.142.24:84/Default.aspx) which is a knowledge base that integrates data from various sources, provides a user-friendly interface for their retrieval, estimates a set of features of interest and computes a confidence score for every candidate protein interaction using a modern computational hybrid methodology.

Although LAPACK is a powerful library its utilization is difficult. JLAPACK, a Java translation obtained automatically from the Fortran LAPACK sources, retains exactly the same difficult to use interface of LAPACK routines. The MTJ library implements an object oriented Java interface to JLAPACK that hides many complicated details. ScalaLab exploits the flexibility of the Scala language to present an even more friendly and convenient interface to the powerful but complicated JLAPACK library. The article describes the interfacing of the low-level JLAPACK routines within the ScalaLab environment. This is performed rather easily by exploiting well suited features of the Scala language. Also, the paper demonstrates the convenience of using JLAPACK routines for linear algebra operations from within ScalaLab.

Protein-Protein Interactions (PPIs) play a very important role in many cellular processes and a variety of experimental approaches have been developed for their identification. These approaches however suffer from high error rates. Recently, computational methods have been employed to assist for the prediction. A common problem with the applied computational methods is that they either result in low predictive performances or produce “black box” classifiers that aren't easily interpretable. In our method we combined Genetic Algorithms and Extended Kalman Filters in order to find the mathematical equation that governs the best classifier. As a result, we are able to construct hypotheses that can explain the complex relationships in the data and biological knowledge can be extrapolated in predictable ways. We tested our hybrid method with a commonly used data set and compared it to previous approaches. We achieved high classification measures (sensitivity: 75.04%, specificity: 84.95%) and outperformed the other methods.

MicroRNAs (miRNAs) are small non coding RNAs that play a significant role in gene regulation. Prediction of microRNA genes is a challenging bioinformatics problem. In our approach we applied an novel classification method, which combines the efficiency and robustness of Support Vector Machines with Genetic Algorithms for feature selection and parameters optimization. We tested our method with commonly used data and feature sets and we achieve higher performance in terms of sensitivity (99,10%), specificity (97,95%) and accuracy (98,68%) than the existing classifiers. Finally, we managed to extract a minimum subset consisted of 7 features that can standalone yield very high classification performance.

Mass spectrometry based methodologies allow us to analyze complex mixtures of proteins from biological samples in order to identify important proteomic patterns and discover novel disease biomarkers. However proteomics data are high-dimensional and complex and need to undergo a number of preprocessing steps so that they can be further analyzed. This procedure involves various steps that interact in complex ways and is an open research area, as it has been shown that it has a big effect on the final results. In this work we propose a novel pipeline method for the preprocessing of MS-based data. Our proposed strategy addresses some of the major problems when dealing with proteomics data, like miscalibration, noise and redundancy. Our method exploits the advantages of using the mean spectrum, ensuring at the same time higher sensitivity. In order to illustrate the utility of our proposed method we experimented on data coming from samples of patients with bladder cancer or benign disease. The results were quite encouraging, as 31 statistically important peaks were identified, some of which are not detected by existing methods.

Abdominal pain in childhood is a common cause of emergency admission in hospital. Its assessment and diagnosis, especially the decision about performing a surgical operation of the abdomen, continues to be a clinical challenge. This study investigates the possibilities of applying state of the art computational intelligence methods for the analysis of abdominal pain data. Specifically, the application of a Genetic Clustering Algorithm and of the Random Forests algorithm (RF) is explored. Clinical appendicitis prediction involves the estimation of at least 15 clinical and laboratory factors (features). The contribution of each factor to the prediction is not known. Thus, the goal of abdominal pain data analysis is not restricted to the classification of the data, but includes the exploration of the underlying data structure. In this study a genetic clustering algorithm is employed for the later task and its performance is compared to a classical K-means clustering approach. For classification purposes, tree methods are frequently used in medical applications since they often reveal simple relationships between variables that can be used to interpret the data. They are however very prone to overfitting problems. Random Forests, applied in this study, is a novel ensemble classifier which builds a number of decision trees to improve the single tree classifier generalization ability. The application of the above mentioned algorithms to real data resulted in very low error rates, (less than 5%), indicating the usefulness of the respective approach. The most informative diagnostic features as proposed by the algorithms are in accordance with known medical expert knowledge. The experimental results furthermore confirmed both, the greater ability of the genetic clustering algorithm to reveal the underlying data patterns as compared to the K-means approach and the effectiveness of RF-based diagnosis as compared to a single decision tree algorithm.

The problem of maximizing the performance of the detection of ischemia episodes is a difficult pattern classification problem. The motivation for developing the supervising network self-organizing map (sNet-SOM) model is to exploit this fact for designing computationally effective solutions both for the particular ischemic detection problem and for other applications that share similar characteristics. Specifically, the sNet-SOM utilizes unsupervised learning for the "simple" regions and supervised for the "difficult" ones in a two stage learning process. The unsupervised learning approach extends and adapts the self-organizing map (SOM) algorithm of Kohonen. The basic SOM is modified with a dynamic expansion process controlled with an entropy based criterion that allows the adaptive formation of the proper SOM structure. This extension proceeds until the total number of training patterns that are mapped to neurons with high entropy reduces to a size manageable numerically with a capable supervised model. The second learning phase has the objective of constructing better decision boundaries at the ambiguous regions. At this phase, a special supervised network is trained for the computationally reduced task of performing the classification at the ambiguous regions only. The utilization of sNet-SOM with supervised learning based on the radial basis functions and support vector machines has resulted in an improved accuracy of ischemia detection especially in the last case. The highly disciplined design of the generalization performance of the support vector machine allows designing the proper model for the number of patterns transferred to the supervised expert.

A computer method for the automatic analysis of specific fluorescence microscopy images, collected with a TUNEL assay procedure applied on human spermatozoa cells, has been developed in this work. In each image, the cells are shown in blue color while in the interior of every cell an area proportional to the degree of the cell DNA fragmentation has been stained in reddish color by the assay procedure. The proposed digital image analysis technique suggests a segmentation method based on the Primary Component Analysis (PCA) applied to the 3-D color space. In this way the shapes in every image are isolated and then a classification technique is used to obtain only the shapes representing solitary spermatozoa, to ignore those presenting cell overlapping and dispose small shapes-scraps belonging to fragments of useless material created during the assay procedure. Finally, for each solitary cell the ratio of the reddish area over the total cell area is estimated.

According to the central dogma of Biology it was commonly accepted that most of the genetic information was transacted by proteins. Recent experimental techniques revealed that the majority of mammalian genomes and other complex organisms are in fact transcribed into non-coding RNAs. Typically, non-coding RNAs are small nucleotide sequences that are not transcribed into proteins and have a profound regulatory role. Present advances in computational biosciences linked their abnormal functionality to many diseases and re-stated the principles of basic therapeutic strategies. The effective identification of non-coding RNAs and their biological role emerges as a new and challenging bioinformatics problem. ncRNAclass ( http://biotools.ceid.upatras.gr/ncrnaclass/ ) is a web platform that allows for efficient computation of a set of features that can describe effectively the broad class of non-coding RNAs. Moreover, it enables the calculation of features that include information about the targeting behavior of miRNAs. The tool operates under a user-friendly interface and its pilot implementation incorporates prediction models for the well-known class of microRNAs and for prediction their mRNA targets. The prediction models are based on two novel evolutionary Machine Learning algorithms that achieve very high classification performance in comparison with existing methods. The platform is also equipped with a data warehouse, with manually curated sequences, that enables fast information retrieval and data mining utilities.

Gene expression datasets are large and complex, having many variables and unknown internal structure. We apply independent component analysis (ICA) to derive a less redundant representation of the expression data. The decomposition produces components with minimal statistical dependence and reveals biologically relevant information. Consequently, to the transformed data, we apply cluster analysis (an important and popular analysis tool for obtaining an initial understanding of the data, usually employed for class discovery). The proposed self-organizing map (SOM)-based clustering algorithm automatically determines the number of ‘natural’ subgroups of the data, being aided at this task by the available prior knowledge of the functional categories of genes. An entropy criterion allows each gene to be assigned to multiple classes, which is closer to the biological representation. These features, however, are not achieved at the cost of the simplicity of the algorithm, since the map grows on a simple grid structure and the learning algorithm remains equal to Kohonen’s one.

Motivation: Currently the most popular approach to analyze genome-wide expression data is clustering. One of the major drawbacks of most of the existing clustering methods is that the number of clusters has to be specified a priori. Furthermore, by using pure unsupervised algorithms prior biological knowledge is totally ignored Moreover, most current tools lack an effective framework for tight integration of unsupervised and supervised learning for the analysis of high-dimensional expression data and only very few multi-class supervised approaches are designed with the provision for effectively utilizing multiple functional class labeling. Results: The paper adapts a novel Self-Organizing map called supervised Network Self-Organized Map (sNet-SOM) to the peculiarities of multi-labeled gene expression data. The sNet-SOM determines adaptively the number of clusters with a dynamic extension process. This process is driven by an inhomogeneous measure that tries to balance unsupervised, supervised and model complexity criteria. Nodes within a rectangular grid are grown at the boundary nodes, weights rippled from the internal nodes towards the outer nodes of the grid, and whole columns inserted within the map The appropriate level of expansion is determined automatically. Multiple sNet-SOM models are constructed dynamically each for a different unsupervised/supervised balance and model selection criteria are used to select the one optimum one. The results indicate that sNet-SOM yields competitive performance to other recently proposed approaches for supervised classification at a significantly reduced computational cost and it provides extensive exploratory analysis potentiality within the analysis framework. Furthermore, it explores simple design decisions that are easier to comprehend and computationally efficient. Availability: The source code of the algorithms presented in the paper can be downloaded from http://heart.med.upatras.gr. The implementation is in Borland C++ Builder 5.0. Contact: severina@heart.med.upatras.gr * To whom correspondence should be addressed.

Since Java is one of the most popular languages in the academic and research community a lot of robust and effective scientific libraries have been developed. However, the utilisation of these libraries is very awkward especially for the average scientist that does not expertise in software development. The study presents the framework that has constructed for the utilisation of Java scientific libraries within the ScalaLab environment. The flexibility and extensibility of the Scala language allows the implementation of simple, coherent and efficient Matlab-like interfaces to those libraries. Moreover, other specialised Java libraries can be exploited much more easily and productively from within ScalaLab with the toolbox import mechanism that this work describes. Additionally, the system offers facilities such as on-line help, code completion, graphical control of the class-path and a specialised text editor with code colouring facilities that greatly facilitate the development of scientific software.

Recently, Self Organizing Maps have been a popular approach to analyze gene expression data. Our paper presents an improved SOM-based algorithm called Supervised Network Self Organizing Map (sNet-SOM), which overcomes the main drawbacks of existing techniques by adaptively determining the number of clusters with a dynamic extension process and integrating unsupervised and supervised learning in an effort to make use of prior knowledge on data. The process is driven by an inhomogeneous measure that balances unsupervised/supervised learning and model complexity criteria. Multiple models are dynamically constructed by the algorithm, each corresponding to an unsupervised/supervised balance, model selection criteria being used to select the optimum one. The design allows us to effectively utilize multiple functional class labeling.