Catálogo de publicaciones - libros

The paper presents a software platform for the management and visualization of mass spectrometry proteomics data. MALDI-TOF and LC-MS spectra can be visualized by considering different parameters or by focusing on portions of spectra. Spectra can also be converted using the XML-based mzData standard for storing or for transmission over the network.

Sequence similarity searching is commonly used to help clarify the biochemical and physiological features of newly discovered genes or proteins. An efficient similarity search relies on the choice of tools and their associated subprograms and numerous parameter settings. This could be very challenging for similarity search users, especially those at the beginner level. To assist researchers in selecting optimal search programs and parameter settings for efficient sequence similarity searches, we have developed a Web-based expert system, Smart Sequence Similarity Search (S4). The system is implemented in Java and Jess scripts, and uses the Jess Expert System as its reasoning core. The expert knowledge provided for a sequence similarity search is represented in the form of decision tree and stored in a XML file. The system also provides interfaces for expert users to improve this knowledge by extending the decision tree. With its capability to continuously improve sequence similarity searches through a decision tree, the Web-based expert system provides a solid advising tool for researchers interested in efficient sequence similarity searches.

The identification of binding sites for transcription factors regulating gene transcription is one of the most important and challenging problems in molecular biology and bioinformatics. Here we present an algorithm that, given a set of promoters from co–regulated genes, identifies over-represented binding sites by using profiles (position specific frequency matrices) defining the sequence binding specificity of known TFs as well as matching statistics on a whole–genome level, bypassing the need of defining matching thresholds and/or the use of homologous sequences. Preliminary tests performed on experimentally validated sequence sets are very promising; moreover, the same algorithm is suitable also for the use with any model of the binding specificity of TFs.

In the last years haplotype reconstruction and haplotype blocks discovery , i.e. , the estimation of patterns of linkage disequilibrium (LD) in the haplotypes, riveted the attention of the computer scientists due to the involved strong computational aspects. Such tasks are usually faced separately; recently, statistical generative techniques permitted to solve them jointly. Following this trend, we propose a generative framework based on hidden Markov processes, equipped with two novel inference strategies. The first strategy estimates finely haplotypes, while the second provides a quantitative measure to estimate LD blocks boundaries. Comparative real data results validate the proposed framework.

Protein fold recognition is an important problem in molecular biology. Machine learning symbolic approaches have been applied to automatically discover local structural signatures and relate these to the concept of fold in SCOP. However, most of these methods cannot handle uncertainty being therefore not able to solve multiple prediction problems. In this paper we present an application of the symbolic-statistical framework PRISM to a multi-class protein fold recognition problem. We compare the proposed approach to a symbolic-only technique and show that the hybrid framework outperforms the symbolic-only one in terms of predictive accuracy in the multiple prediction problem.

The application of genome-wide approaches to the molecular characterization of cancer was investigated, identifying footprints that can potentially assist in the subclassification of tumors in order to contribute to diagnosis and clinical management of patients. High resolution DNA copy number analysis by single nucleotide polymorphism mapping array technology has been widely applied to study copy number aberrations and to distinguish among different loss of heterozigosity mechanisms associated with or without copy number changes in tumors. However, assessment of statistically significant common aberrations across the whole data set or a subset of tumor samples is still an open problem. Therefore, we adapted the recently developed STAC algorithm, previously applied to comparative genomic hybridization data, to identify common copy number aberrations in renal carcinoma samples using Affymetrix 100K SNP arrays. SNP copy number data were processed by a homebrew pipeline implemented in R and analyzed using STAC.

The systematic integration of expression profiles and other types of gene information, such as copy number, chromosomal localization, and sequence characteristics, still represents a challenge in the genomic arena. In particular, the integrative analysis of genomic and transcriptional data in context of the physical location of genes in a genome appears promising in detecting chromosomal regions with structural and transcriptional imbalances often characterizing cancer. A computational framework based on locally adaptive statistical procedures (Global Smoothing Copy Number, GLSCN, and Locally Adaptive Statistical Procedure, LAP), which incorporate genomic and transcriptional data with structural information for the identification of imbalanced chromosomal regions, is described. Both GLSCN and LAP accounts for variations in the distance between genes and in gene density by smoothing standard statistics on gene position before testing the significance of copy number and gene expression signals. The application of GLSCN and LAP to the integrative analysis of a human metastatic clear cell renal carcinoma cell line (Caki-1) allowed identifying chromosomal regions that are directly involved in known chromosomal aberrations characteristic of tumors.