MOTIVATION: Recent large-scale chromatin states mapping efforts have revealed characteristic chromatin modification signatures for various types of functional DNA elements. Given the important influence of chromatin states on gene regulation and the rapid accumulation of genome-wide chromatin modification data, there is a pressing need for computational methods to analyze these data in order to identify functional DNA elements. However, existing computational tools do not exploit data transformation and feature extraction as a means to achieve a more accurate prediction. RESULTS: We introduce a new computational framework for identifying functional DNA elements using chromatin signatures. The framework consists of a data transformation and a feature extraction step followed by a classification step using time-delay neural network. We implemented our framework in a software tool CSI-ANN (chromatin signature identification by artificial neural network). When applied to predict transcriptional enhancers in the ENCODE region, CSI-ANN achieved a 65.5% sensitivity and 66.3% positive predictive value, a 5.9% and 11.6% improvement, respectively, over the previously best approach. AVAILABILITY AND IMPLEMENTATION: CSI-ANN is implemented in Matlab. The source code is freely available at http://www.medicine.uiowa.edu/Labs/tan/CSIANNsoft.zip CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary Materials are available at Bioinformatics online.

The ‘Handbook of Bioinspired Algorithms and Applications’ is a unifying compilation of chapters that discuss bioinspired algorithms. These are algorithms that have been conceived using cues from nature, such as, evolution, insect swarming, food foraging and brain neurons. The book presents a broad range of solved problems and cases studied using bioinspired paradigms. The book is well organized, each chapter being condensed but detailed enough to understand the topics for those researchers who have little or no background in bioinspired algorithms. At the end of each chapter plenty of carefully selected references are provided for further information. The book also contains a global index for easy searching. This 618-page book is divided in two sections: Models and Paradigms and Application Domains with 38 contributions. The first section consists of 11 chapters and the second by 25 chapters. The first section of the book provides the background needed to understand bioinspired algorithms. It discusses the following: evolutionary algorithms (EA), genetic algorithms (GA), genetic programming (GP), particle swarm optimization (PSO) and ant colony optimization (ACO). There is a chapter dedicated to neural networks (NN), which are universal approximators, and another for cellular automata (CA). A chapter on DNA computing is also included; the algorithms are based on Darwin's evolutionary theory where operators are selection, crossover, mutation or variants of these. For instance, the velocity update of the PSO algorithm is considered a weak mutation operator from an evolutionary algorithm's perspective. Chapters describe theory and design decisions, pseudo-codes, parameter selection, parallelization and performance comparisons between different algorithms. The second section of the book is introduced with a few chapters discussing GA implementation for scheduling task problems, network traffic problems, network configuration and medical diagnosis using different imaging techniques e.g., fMRI and X-rays. There are chapters that discuss applications of cellular automata in scheduling problems and pattern formation. Later, solutions are presented for routing algorithms, partition problems and data clustering using swarming algorithms, i.e., PSO and ACO. After that, the next chapters discuss applications using NNs. An intrusion detection system for mobile phone users and synthesis of multiple-value circuit problems were approached using NNs. Additionally, an extensive chapter discusses the computational capacity of NNs using discrete multiple-valued multiple-threshold neurons. An interesting chapter presents advanced evolutionary algorithms to train NNs. Results show that the evolutionary strategy yields more satisfactory results than some problem-specific algorithms. Later, bioinspired algorithms for data mining (also known as knowledge discovery in databases, KDD) are introduced. Different EA- and swarming-based algorithms are presented along with experiments and with a performance analysis. Another chapter presents a GA-based data mining routine applied to microarray analysis. A chapter, less related to the field of bioinformatics and computational biology, gives to the reader an insight into how problem solutions are encoded in EAs, and presents several applications of EAs to solve different electrical engineering problems such as integrate circuit design and universal motor design. The book continues with more applications of EAs to partitioning problems. A relatively extensive chapter in population-learning algorithms (PLA), a type of combinatorial optimization algorithm that emulates social processes and applications of these such as permutation scheduling are presented. After this, an exciting chapter studies a biological model that emulates the evolution of micro-electro mechanical system (MEMS) technology that can be used as a guiding framework for their study. The last three chapters discuss parallel and distributed frameworks for cooperative meta-heuristic algorithms and EAs and an application to the coloring problem. Finally, parallel multi-objective meta-heuristic algorithms are discussed and implemented for a peer-to-peer problem. In conclusion, I recommend this book as a reference book or text book for a special topic class. It provides a general insight for those researchers who are not familiar with evolutionary algorithms and neural networks. Case studies and solved problems provide a clear idea of the kind of problems to which EAs can be applied and how these are encoded in EAs.

In this paper, we describe a general-purpose, systematic algorithm, consisting of a genetic programming module and a k-nearest neighbor classifier to automatically create artificial features-features that are computer-crafted and may not have a known physical meaning-directly from the reconstructed state-space trajectories of the EEG signals that reveal patterns indicative of epileptic seizure onset. The algorithm was evaluated in three patients and validation experiments were carried out using 267.6 hours of EEG recordings. The results with the artificial features compare favorably with previous benchmark work that used a handcrafted feature.