Pattern Recognition in Bioinformatics

Pattern Recognition in Bioinformatics

Pattern Recognition Letters 31 (2010) 2071–2072 Contents lists available at ScienceDirect Pattern Recognition Letters journal homepage: www.elsevier...

122KB Sizes 2 Downloads 79 Views

Pattern Recognition Letters 31 (2010) 2071–2072

Contents lists available at ScienceDirect

Pattern Recognition Letters journal homepage:


Pattern Recognition in Bioinformatics

With the advent of high speed computers, in-silico studies on biological patterns in recent years have been significantly impacted by the pattern recognition techniques. In this special issue, ‘Pattern Recognition in Bioinformatics’, we present various sophisticated algorithms for a wide range of pattern recognition problems from the world of complex biological systems, whether these are specific sequence signatures – motifs that stand out in discovering its partner – or substructures in an interaction network that determines an organisms’ response to external stimuli. The 12 high-quality articles included in this special issue are essentially based on significant extensions of the selected papers presented at the Third International Conference on Pattern Recognition in Bioinformatics (PRIB 2008) held in Melbourne, Australia. All these selected papers for special issue have again undergone a thorough review by at least three reviewers who are experts in the field. The fresh review process was followed to ensure that the papers met the high standards of scientific and technical merit of the Pattern Recognition Letters journal. The issue is broadly divided into three sections of four papers each, namely (1) Section 1: Interaction Networks and Feature-based Predictions (2) Section 2: Microarray and Transcription Data Analysis (3) Section 3: Sequence Analysis and Motif Discovery In Section 1, on ‘Interaction Networks and Feature-based Predictions’, the first paper by Haiying Wang et al. describes a method to integrate similarities between proteins derived from Gene Ontology based annotations and shows how they can be used to construct a protein–protein interaction network. The authors present a framework for the probabilistic combination of semantic similarity knowledge extracted from well-known Gene Ontology hierarchies for the analysis of protein–protein interaction networks and successfully demonstrate its application to yeast genome. The second paper, by Pavol Jancura and Elena Marchiori, proposes an algorithm to divide a protein–protein interaction network into modules for efficient comparison and alignment. The method divides a protein–protein interaction network into subgraphs, which serve as the basic units of comparison between networks. In real-life models, lower-level network units are provided by binary interaction prediction. In this regard, Ashish Anand et al. describe an algorithm to predict interactions between transcription factors and their targets using feature-based description of the two systems. Feature selection procedures have been developed to identify family specific signatures of TF-target complexes. 0167-8655/$ - see front matter Ó 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.patrec.2010.06.001

In the next paper, an even higher resolution prediction i.e. binding sites (in this case calcium-binding sites) has been reported by Jeremy Horst and Ram Samudrala. This is an extension of a well-known technique of predicting one-dimensional features of proteins, but goes further by taking structure-derived features as the inputs and identifies specific properties that allow certain residues to cooperatively identify calcium ions. In Section 2, on ‘Microarray and Transcription Data Analysis’, Alioune Ngom et al. report a method to select an oligonucleotide probe in selection in Microarray experiments. In this paper, the authors develop heuristics, which are guided by a set of selection functions defined over a probe set. These heuristics are used to decide at each moment which probes are the best to be included in, or excluded from, a candidate solution. In the next paper, Yvonne Pittelkow and Susan Wilson come up with an algorithm to find transcriptional regulation patterns in cell-cycle, making an extensive analysis of experimental data and applying an iterative procedure. Dynamic pattern discovery in gene regulation is obtained by introducing time-series parameters into the model, a task which has been achieved by Iti Chaturvedi and Jagath Rajapakse. A combination of dynamic Bayesian networks and hidden Markovian models is used to model dynamic behavior of expression data. This section ends with a paper on the case study of cancer etiology performed by Macintyre et al., who specifically use Gene Ontology patterns for their study. This paper specifically demonstrates how functionally important genes with weakly correlated expressions can be discovered using Gene Ontology information. Finally, the last section on ‘Sequence Analysis and Motif Discovery’ is dedicated to the classical problem of sequence analysis and motif discovery. A critical review of pattern discovery algorithms using Maximum likelihood has been performed by Chengpeng Bi. The paper develops a framework, which unifies a suite of motiffinding algorithms by maximizing the same function. This enables a systematic comparison of different optimization schemes as well as provides a practical guidance on using these techniques. This paper is followed by a novel and simplified algorithm to identify sequence repeats by Radhakrishnan Sabarinathan et al. In the next paper, a low level processing unit manipulation for motif discovery is provided by Yongchao Liu et al. In this paper, the authors present a highly parallel formulation and implementation of the MEME motif discovery algorithm using the CUDA programming model. This system speeds up sequence motif discovery manifold. Finally, a case study on specific Leshimania spp. genomic system has been reported by Michely Diniz et al. A combination of Multi-relational data mining (MRDM), Hidden Markov models (HMMs) and Viterbi algorithm (VA) has been applied to the genome databases of


Editorial / Pattern Recognition Letters 31 (2010) 2071–2072

pathogenic protozoa Leishmania and the integrated approach has been compared with heuristics based approach largely employed for motif discovery. We would like to acknowledge the authors for their excellent contributions and also for their patience during the long review and production process. The Guest Editors would also like to thank the expert reviewers who were generous enough to devote their time in giving excellent feedback and constructive criticisms. The reviews helped authors in improving their manuscripts and editors to make scientifically sound judgments on the quality of articles. The Guest Editors would also like to thank the PATREC Special Issues Editor-in-Chief, Gabriella Sanniti di Baja, for all her contributions and support and also Gary Anderton, the Journal Manager, for efficiently monitoring the entire review and editorial activities. There is no doubt they led the development of this special issue

with great sincerity and competence and, without their help, we would have not been able to complete this task. We hope you find the articles to be interesting and motivating. Guest Editors Shandar Ahmad Madhu Chetty Bertil Schmidt Tel.: +81 72 641 9848; fax: +81 72 641 9812 E-mail addresses: [email protected] (S. Ahmad), [email protected] (M. Chetty), [email protected] (B. Schmidt). Available online 4 June 2010