José Luís Oliveira

Biological and Medical Data Analysis

Medical Databases and Information Systems.- Application of Three-Level Handprinted Documents Recognition in Medical Information Systems.- Data Management and Visualization Issues in a Fully Digital Echocardiography Laboratory.- A Framework Based on Web Services and Grid Technologies for Medical Image Registration.- Biomedical Image Processing Integration Through INBIOMED: A Web Services-Based Platform.- The Ontological Lens: Zooming in and out from Genomic to Clinical Level.- Data Analysis and Image Processing.- Dynamics of Vertebral Column Observed by Stereovision and Recurrent Neural Network Model.- Endocardial Tracking in Contrast Echocardiography Using Optical Flow.- Unfolding of Virtual Endoscopy Using Ray-Template.- Knowledge Discovery and Data Mining.- Integration of Genetic and Medical Information Through a Web Crawler System.- Vertical Integration of Bioinformatics Tools and Information Processing on Analysis Outcome.- A Grid Infrastructure for Text Mining of Full Text Articles and Creation of a Knowledge Base of Gene Relations.- Prediction of the Performance of Human Liver Cell Bioreactors by Donor Organ Data.- A Bioinformatic Approach to Epigenetic Susceptibility in Non-disjunctional Diseases.- Foreseeing Promising Bio-medical Findings for Effective Applications of Data Mining.- Statistical Methods and Tools for Biomedical Data Analysis.- Hybridizing Sparse Component Analysis with Genetic Algorithms for Blind Source Separation.- Hardware Approach to the Artificial Hand Control Algorithm Realization.- Improving the Therapeutic Performance of a Medical Bayesian Network Using Noisy Threshold Models.- SVM Detection of Premature Ectopic Excitations Based on Modified PCA.- Decision Support Systems.- A Text Corpora-Based Estimation of the Familiarity of Health Terminology.- On Sample Size and Classification Accuracy: A Performance Comparison.- Influenza Forecast: Comparison of Case-Based Reasoning and Statistical Methods.- Tumor Classification from Gene Expression Data: A Coding-Based Multiclass Learning Approach.- Boosted Decision Trees for Diagnosis Type of Hypertension.- Markov Chains Pattern Recognition Approach Applied to the Medical Diagnosis Tasks.- Computer-Aided Sequential Diagnosis Using Fuzzy Relations - Comparative Analysis of Methods.- Collaborative Systems in Biomedical Informatics.- Service Oriented Architecture for Biomedical Collaborative Research.- Simultaneous Scheduling of Replication and Computation for Bioinformatic Applications on the Grid.- The INFOBIOMED Network of Excellence: Developments for Facilitating Training and Mobility.- Bioinformatics: Computational Models.- Using Treemaps to Visualize Phylogenetic Trees.- An Ontological Approach to Represent Molecular Structure Information.- Focal Activity in Simulated LQT2 Models at Rapid Ventricular Pacing: Analysis of Cardiac Electrical Activity Using Grid-Based Computation.- Bioinformatics: Structural Analysis.- Extracting Molecular Diversity Between Populations Through Sequence Alignments.- Detection of Hydrophobic Clusters in Molecular Dynamics Protein Unfolding Simulations Using Association Rules.- Protein Secondary Structure Classifiers Fusion Using OWA.- Efficient Computation of Fitness Function by Pruning in Hydrophobic-Hydrophilic Model.- Evaluation of Fuzzy Measures in Profile Hidden Markov Models for Protein Sequences.- Bioinformatics: Microarray Data Analysis.- Relevance, Redundancy and Differential Prioritization in Feature Selection for Multiclass Gene Expression Data.- Gene Selection and Classification of Human Lymphoma from Microarray Data.- Microarray Data Analysis and Management in Colorectal Cancer.

Large Scale Comparative Codon-Pair Context Analysis Unveils General Rules that Fine-Tune Evolution of mRNA Primary Structure

Background Codon usage and codon-pair context are important gene primary structure features that influence mRNA decoding fidelity. In order to identify general rules that shape codon-pair context and minimize mRNA decoding error, we have carried out a large scale comparative codon-pair context analysis of 119 fully sequenced genomes. Methodologies/Principal Findings We have developed mathematical and software tools for large scale comparative codon-pair context analysis. These methodologies unveiled general and species specific codon-pair context rules that govern evolution of mRNAs in the 3 domains of life. We show that evolution of bacterial and archeal mRNA primary structure is mainly dependent on constraints imposed by the translational machinery, while in eukaryotes DNA methylation and tri-nucleotide repeats impose strong biases on codon-pair context. Conclusions The data highlight fundamental differences between prokaryotic and eukaryotic mRNA decoding rules, which are partially independent of codon usage.

Comparative context analysis of codon pairs on an ORFeome scale

Codon context is an important feature of gene primary structure that modulates mRNA decoding accuracy. We have developed an analytical software package and a graphical interface for comparative codon context analysis of all the open reading frames in a genome (the ORFeome). Using the complete ORFeome sequences of Saccharomyces cerevisiae, Schizosaccharomyces pombe, Candida albicans and Escherichia coli, we show that this methodology permits large-scale codon context comparisons and provides new insight on the rules that govern the evolution of codon-pair context.

Parallel DNA pyrosequencing unveils new zebrafish microRNAs

BackgroundMicroRNAs (miRNAs) are a new class of small RNAs of approximately 22 nucleotides in length that control eukaryotic gene expression by fine tuning mRNA translation. They regulate a wide variety of biological processes, namely developmental timing, cell differentiation, cell proliferation, immune response and infection. For this reason, their identification is essential to understand eukaryotic biology. Their small size, low abundance and high instability complicated early identification, however cloning/Sanger sequencing and new generation genome sequencing approaches overcame most technical hurdles and are being used for rapid miRNA identification in many eukaryotes.ResultsWe have applied 454 DNA pyrosequencing technology to miRNA discovery in zebrafish (Danio rerio). For this, a series of cDNA libraries were prepared from miRNAs isolated at different embryonic time points and from fully developed organs. Each cDNA library was tagged with specific sequences and was sequenced using the Roche FLX genome sequencer. This approach retrieved 90% of the 192 miRNAs previously identified by cloning/Sanger sequencing and bioinformatics. Twenty five novel miRNAs were predicted, 107 miRNA star sequences and also 41 candidate miRNA targets were identified. A miRNA expression profile built on the basis of pyrosequencing read numbers showed high expression of most miRNAs throughout zebrafish development and identified tissue specific miRNAs.ConclusionThis study increases the number of zebrafish miRNAs from 192 to 217 and demonstrates that a single DNA mini-chip pyrosequencing run is effective in miRNA identification in zebrafish. This methodology also produced sufficient information to elucidate miRNA expression patterns during development and in differentiated organs. Moreover, some zebrafish miRNA star sequences were more abundant than their corresponding miRNAs, suggesting a functional role for the former in gene expression control in this vertebrate model organism.

Gimli: open source and high-performance biomedical name recognition

BackgroundAutomatic recognition of biomedical names is an essential task in biomedical information extraction, presenting several complex and unsolved challenges. In recent years, various solutions have been implemented to tackle this problem. However, limitations regarding system characteristics, customization and usability still hinder their wider application outside text mining research.ResultsWe present Gimli, an open-source, state-of-the-art tool for automatic recognition of biomedical names. Gimli includes an extended set of implemented and user-selectable features, such as orthographic, morphological, linguistic-based, conjunctions and dictionary-based. A simple and fast method to combine different trained models is also provided. Gimli achieves an F-measure of 87.17% on GENETAG and 72.23% on JNLPBA corpus, significantly outperforming existing open-source solutions.ConclusionsGimli is an off-the-shelf, ready to use tool for named-entity recognition, providing trained and optimized models for recognition of biomedical entities from scientific text. It can be used as a command line tool, offering full functionality, including training of new models and customization of the feature set and model parameters through a configuration file. Advanced users can integrate Gimli in their text mining workflows through the provided library, and extend or adapt its functionalities. Based on the underlying system characteristics and functionality, both for final users and developers, and on the reported performance results, we believe that Gimli is a state-of-the-art solution for biomedical NER, contributing to faster and better research in the field. Gimli is freely available at http://bioinformatics.ua.pt/gimli.

A PACS archive architecture supported on cloud services

PurposeDiagnostic imaging procedures have continuously increased over the last decade and this trend may continue in coming years, creating a great impact on storage and retrieval capabilities of current PACS. Moreover, many smaller centers do not have financial resources or requirements that justify the acquisition of a traditional infrastructure. Alternative solutions, such as cloud computing, may help address this emerging need.MethodsA tremendous amount of ubiquitous computational power, such as that provided by Google and Amazon, are used every day as a normal commodity. Taking advantage of this new paradigm, an architecture for a Cloud-based PACS archive that provides data privacy, integrity, and availability is proposed. The solution is independent from the cloud provider and the core modules were successfully instantiated in examples of two cloud computing providers. Operational metrics for several medical imaging modalities were tabulated and compared for Google Storage, Amazon S3, and LAN PACS.ResultsA PACS-as-a-Service archive that provides storage of medical studies using the Cloud was developed. The results show that the solution is robust and that it is possible to store, query, and retrieve all desired studies in a similar way as in a local PACS approach.ConclusionCloud computing is an emerging solution that promises high scalability of infrastructures, software, and applications, according to a “pay-as-you-go” business model. The presented architecture uses the cloud to setup medical data repositories and can have a significant impact on healthcare institutions by reducing IT infrastructures.

Dicoogle - an open source peer-to-peer PACS.

Picture Archiving and Communication Systems (PACS) have been widely deployed in healthcare institutions, and they now constitute a normal commodity for practitioners. However, its installation, maintenance, and utilization are still a burden due to their heavy structures, typically supported by centralized computational solutions. In this paper, we present Dicoogle, a PACS archive supported by a document-based indexing system and by peer-to-peer (P2P) protocols. Replacing the traditional database storage (RDBMS) by a documental organization permits gathering and indexing data from file-based repositories, which allows searching the archive through free text queries. As a direct result of this strategy, more information can be extracted from medical imaging repositories, which clearly increases flexibility when compared with current query and retrieval DICOM services. The inclusion of P2P features allows PACS internetworking without the need for a central management framework. Moreover, Dicoogle is easy to install, manage, and use, and it maintains full interoperability with standard DICOM services.

Twitter: A Good Place to Detect Health Conditions

With the proliferation of social networks and blogs, the Internet is increasingly being used to disseminate personal health information rather than just as a source of information. In this paper we exploit the wealth of user-generated data, available through the micro-blogging service Twitter, to estimate and track the incidence of health conditions in society. The method is based on two stages: we start by extracting possibly relevant tweets using a set of specially crafted regular expressions, and then classify these initial messages using machine learning methods. Furthermore, we selected relevant features to improve the results and the execution times. To test the method, we considered four health states or conditions, namely flu, depression, pregnancy and eating disorders, and two locations, Portugal and Spain. We present the results obtained and demonstrate that the detection results and the performance of the method are improved after feature selection. The results are promising, with areas under the receiver operating characteristic curve between 0.7 and 0.9, and f-measure values around 0.8 and 0.9. This fact indicates that such approach provides a feasible solution for measuring and tracking the evolution of health states within the society.

BeCAS: biomedical concept recognition services and visualization

SUMMARY The continuous growth of the biomedical scientific literature has been motivating the development of text-mining tools able to efficiently process all this information. Although numerous domain-specific solutions are available, there is no web-based concept-recognition system that combines the ability to select multiple concept types to annotate, to reference external databases and to automatically annotate nested and intercepted concepts. BeCAS, the Biomedical Concept Annotation System, is an API for biomedical concept identification and a web-based tool that addresses these limitations. MEDLINE abstracts or free text can be annotated directly in the web interface, where identified concepts are enriched with links to reference databases. Using its customizable widget, it can also be used to augment external web pages with concept highlighting features. Furthermore, all text-processing and annotation features are made available through an HTTP REST API, allowing integration in any text-processing pipeline. AVAILABILITY BeCAS is freely available for non-commercial use at http://bioinformatics.ua.pt/becas. CONTACTS tiago.nunes@ua.pt or jlo@ua.pt.

On the parameter optimization of Support Vector Machines for binary classification

Classifying biological data is a common task in the biomedical context. Predicting the class of new, unknown information allows researchers to gain insight and make decisions based on the available data. Also, using classification methods often implies choosing the best parameters to obtain optimal class separation, and the number of parameters might be large in biological datasets. Support Vector Machines provide a well-established and powerful classification method to analyse data and find the minimal-risk separation between different classes. Finding that separation strongly depends on the available feature set and the tuning of hyper-parameters. Techniques for feature selection and SVM parameters optimization are known to improve classification accuracy, and its literature is extensive. In this paper we review the strategies that are used to improve the classification performance of SVMs and perform our own experimentation to study the influence of features and hyper-parameters in the optimization process, using several known kernels.