SiBIC: A Web Server for Generating Gene Set Networks Based on Biclusters Obtained by Maximal Frequent Itemset Mining

Detecting biclusters from expression data is useful, since biclusters are coexpressed genes under only part of all given experimental conditions. We present a software called SiBIC, which from a given expression dataset, first exhaustively enumerates biclusters, which are then merged into rather independent biclusters, which finally are used to generate gene set networks, in which a gene set assigned to one node has coexpressed genes. We evaluated each step of this procedure: 1) significance of the generated biclusters biologically and statistically, 2) biological quality of merged biclusters, and 3) biological significance of gene set networks. We emphasize that gene set networks, in which nodes are not genes but gene sets, can be more compact than usual gene networks, meaning that gene set networks are more comprehensible. SiBIC is available at http://utrecht.kuicr.kyoto-u.ac.jp:8080/miami/faces/index.jsp.     

Gene Network and Database for Genes of Wheat’s Resistance to Pathogenic Fungi

An overview describing a gene network that controls the formation of plant responses to diseases caused by pathogenic fungi (http://wwwmgs.bionet.nsc.ru/mgs/gnw/genenet//viewer/Plant%20fungus%20pathogen.html) is presented. The gene network represents the coordinated interactions of genes, proteins, and regulatory molecules, including integrated defense mechanisms that prevent the development of infection, localize the lesion, and minimize damage. The gene network was reconstructed on the basis of literature data, and the elements of the gene network were associated with the records of the PGR database (Pathogenesis-Related Genes, http://srs6.bionet.nsc.ru/srs6bin/cgi-bin/wgetz?-page+top+-newId), where information on plant genes resistant to pathogenic fungi is accumulated. Reconstruction of the gene network allows us to formalize, visualize, and systematize possible mechanisms for the response of plant cells to fungal infection, which may be useful for the planning of experiments and interpretation of experimental data in this field of science.     

Prediction of eukaryotic mRNA translational properties.

MOTIVATION: It is well known that eukaryotic mRNAs are translated at different levels depending on their sequence characteristics. Evaluation of mRNA translatability is of importance in prediction of the gene expression pattern by computer methods and to improve the recognition of mRNAs within cloned nucleotide sequences. It may also be used in biotechnological experiments to optimize the expression of foreign genes in transgenic organisms. RESULTS: The sets of 5' untranslated region characteristics, significantly different between mRNAs encoding abundant and scarce polypeptides, were determined for mammals, dicot plants and monocot plants, and collected in the LEADER_RNA database. Computer tools for the prediction of mRNA translatability are presented. AVAILABILITY: Programs for mRNA translatability prediction are available at http://wwwmgs.bionet.nsc. ru/programs/acts2/mo_mRNA.htm (for monocots), http://wwwmgs.bionet. nsc.ru/programs/acts2/di_mRNA.htm (for dicots) and http://wwwmgs. bionet.nsc.ru/programs/acts2/ma_mRNA.htm (for mammals). The LEADER_RNA database may be accessed at: http://wwwmgs.bionet.nsc. ru/systems/LeaderRNA/.     

New method to predict conserved RNA structures

RNA secondary structure prediction is one of the classic problems of bioinformatics. The most efficient approaches to solving this problem are based on comparative analysis. As a rule, multiple RNA sequence alignment and subsequent determination of a common secondary structure are used. A new algorithm was developed to obviate the need for preliminary multiple sequence alignment. The algorithm is based on a multilevel MEME-like iterative search for a generalized profile. The search for common blocks in RNA sequences is carried out at the first level. Then the algorithm refines the chains consisting of these blocks. Finally, the search for sets of common helices, matched with alignment blocks, is carried out. The algorithm was tested with a tRNA set containing additional junk sequences and with RFN riboswitches. The algorithm is available at http://bioinf.fbb.msu.ru/RNAAlign.     

A method for prediction of conserved RNA secondary structures

The RNA secondary structure prediction is a classical problem in bioinformatics. The most efficient approach to this problem is based on the idea of a comparative analysis. In this approach the algorithms utilize multiple alignment of the RNA sequences and find common RNA structure. This paper describes a new algorithm for this task. This algorithm does not require predefined multiple alignment. The main idea of the algorithm is based on MEME-like iterative searching of abstract profile on different levels. On the first level the algorithm searches the common blocks in the RNA sequences and creates chain of this blocks. On the next step the algorithm refines the chain of common blocks. On the last stage the algorithm searches sets of common helices that have consistent locations relative to common blocks. The algorithm was tested on sets of tRNA with a subset of junk sequences and on RFN riboswitches. The algorithm is implemented as a web server (http://bioinf.fbb.msu.ru/RNAAlign/).     

CRCView: a web server for analyzing and visualizing microarray gene expression data using model-based clustering

CRCView is a user-friendly point-and-click web server for analyzing and visualizing microarray gene expression data using a Dirichlet process mixture model-based clustering algorithm. CRCView is designed to clustering genes based on their expression profiles. It allows flexible input data format, rich graphical illustration as well as integrated GO term based annotation/interpretation of clustering results. Availability: http://helab.bioinformatics.med.umich.edu/crcview/.     

Dial-a-mutant: web-based knockout collections for model organisms

Although much of eukaryotic cell biology is highly conserved, the most biomedically relevant organisms are often the most difficult or expensive to study. Non-vertebrate model organisms offer a practical and accessible alternative. Large collections of loss-of-function mutants have now been generated in the nematode Caenorhabditis elegans and the angiosperm Arabidopsis thaliana, which can be easily accessed via the Internet. The sites reviewed in the present article include those of the C. elegans Knockout Consortium (http://celeganskoconsortium.omrf.org/) and the C. elegans National Bioresource Project of Japan (http://shigen.lab.nig.ac.jp/c.elegans/index.jsp), and also the NemaGENETAG project (http://elegans.imbb.forth.gr/nemagenetag/) for the former and T-DNA Express (http://signal.salk.edu/cgi-bin/tdnaexpress) for the latter. These sites allow one to easily identify and request C. elegans strains bearing mutations in approximately half of predicted genes and for essentially every Arabidopsis gene. These tools greatly enhance the ability of non-specialists to conduct studies of gene function in vivo in these model species.     

A database for management of gene expression data in situ

MOTIVATION: To create a spatiotemporal atlas of Drosophila segmentation gene expression at cellular resolution. RESULTS: The expression of segmentation genes plays a crucial role in the establishment of the Drosophila body plan. Using the IBM DB2 Relational Database Management System we have designed and implemented the FlyEx database. FlyEx contains 2832 images of 14 segmentation gene expression patterns obtained from 954 embryos and 2,073,662 quantitative data records. The averaged data is available for most of segmentation genes at eight time points. FlyEx supports operations on images of gene expression patterns. The database can be used to examine the quality of data, analyze the dynamics of formation of segmentation gene expression domains, as well as estimate the variability of gene expression patterns. We also provide the capability to download data of interest. AVAILABILITY: http://urchin.spbcas.ru/flyex, http://flyex.ams.sunysb.edu/flyex     

Finding edging genes from microarray data

MOTIVATION: A set of genes and their gene expression levels are used to classify disease and normal tissues. Due to the massive number of genes in microarray, there are a large number of edges to divide different classes of genes in microarray space. The edging genes (EGs) can be co-regulated genes, they can also be on the same pathway or deregulated by the same non-coding genes, such as siRNA or miRNA. Every gene in EGs is vital for identifying a tissue's class. The changing in one EG's gene expression may cause a tissue alteration from normal to disease and vice versa. Finding EGs is of biological importance. In this work, we propose an algorithm to effectively find these EGs. RESULT: We tested our algorithm with five microarray datasets. The results are compared with the border-based algorithm which was used to find gene groups and subsequently divide different classes of tissues. Our algorithm finds a significantly larger amount of EGs than does the border-based algorithm. As our algorithm prunes irrelevant patterns at earlier stages, time and space complexities are much less prevalent than in the border-based algorithm. AVAILABILITY: The algorithm proposed is implemented in C++ on Linux platform. The EGs in five microarray datasets are calculated. The preprocessed datasets and the discovered EGs are available at http://www3.it.deakin.edu.au/~phoebe/microarray.html.     

Computational method for predicting protein functional sites with the use of specificity determinants

The currently available body of decoded amino acid sequences of various proteins exceeds manifold the experimental capabilities of their functional annotation. Therefore, in silico annotation using bioinformatics methods becomes increasingly important. Such annotation is actually a prediction; however, this can be an important starting point for further laboratory research. This work describes a new method for predicting functionally important protein sites, SDPsite, on the basis of identification of specificity determinants. The algorithm proposed utilizes a protein family aglinment and a phylogenetic tree to predict the conserved positions and specificity determinants, map them onto the protein structure, and search for clusters of the predicted positions. Comparison of the resulting predictions with experimental data and published predictions of functional sites by other methods demonstrates that the results of SDPsite agree well with experimental data and exceed the results obtained with the majority of previous methods. SDPsite is publicly available at http://bioinf.fbb.msu.ru/SDPsite.     

Gene structure prediction from consensus spliced alignment of multiple ESTs matching the same genomic locus

MOTIVATION: Accurate gene structure annotation is a challenging computational problem in genomics. The best results are achieved with spliced alignment of full-length cDNAs or multiple expressed sequence tags (ESTs) with sufficient overlap to cover the entire gene. For most species, cDNA and EST collections are far from comprehensive. We sought to overcome this bottleneck by exploring the possibility of using combined EST resources from fairly diverged species that still share a common gene space. Previous spliced alignment tools were found inadequate for this task because they rely on very high sequence similarity between the ESTs and the genomic DNA. RESULTS: We have developed a computer program, GeneSeqer, which is capable of aligning thousands of ESTs with a long genomic sequence in a reasonable amount of time. The algorithm is uniquely designed to tolerate a high percentage of mismatches and insertions or deletions in the EST relative to the genomic template. This feature allows use of non-cognate ESTs for gene structure prediction, including ESTs derived from duplicated genes and homologous genes from related species. The increased gene prediction sensitivity results in part from novel splice site prediction models that are also available as a stand-alone splice site prediction tool. We assessed GeneSeqer performance relative to a standard Arabidopsis thaliana gene set and demonstrate its utility for plant genome annotation. In particular, we propose that this method provides a timely tool for the annotation of the rice genome, using abundant ESTs from other cereals and plants. AVAILABILITY: The source code is available for download at http://bioinformatics.iastate.edu/bioinformatics2go/gs/download.html. Web servers for Arabidopsis and other plant species are accessible at http://www.plantgdb.org/cgi-bin/AtGeneSeqer.cgi and http://www.plantgdb.org/cgi-bin/GeneSeqer.cgi, respectively. For non-plant species, use http://bioinformatics.iastate.edu/cgi-bin/gs.cgi. The splice site prediction tool (SplicePredictor) is distributed with the GeneSeqer code. A SplicePredictor web server is available at http://bioinformatics.iastate.edu/cgi-bin/sp.cgi     

Mining gene expression data using a novel approach based on hidden Markov models

In this work we have developed a new framework for microarray gene expression data analysis. This framework is based on hidden Markov models. We have benchmarked the performance of this probability model-based clustering algorithm on several gene expression datasets for which external evaluation criteria were available. The results showed that this approach could produce clusters of quality comparable to two prevalent clustering algorithms, but with the major advantage of determining the number of clusters. We have also applied this algorithm to analyze published data of yeast cell cycle gene expression and found it able to successfully dig out biologically meaningful gene groups. In addition, this algorithm can also find correlation between different functional groups and distinguish between function genes and regulation genes, which is helpful to construct a network describing particular biological associations. Currently, this method is limited to time series data. Supplementary materials are available at http://www.bioinfo.tsinghua.edu.cn/~rich/hmmgep_supp/.     

Bioinformatics Approach to Evaluate Differential Gene Expression of M1/M2 Macrophage Phenotypes and Antioxidant Genes in Atherosclerosis

Atherosclerosis is a pro-inflammatory process intrinsically related to systemic redox impairments. Macrophages play a major role on disease development. The specific involvement of classically activated, M1 (pro-inflammatory), or the alternatively activated, M2 (anti-inflammatory), on plaque formation and disease progression are still not established. Thus, based on meta-data analysis of public micro-array datasets, we compared differential gene expression levels of the human antioxidant genes (HAG) and M1/M2 genes between early and advanced human atherosclerotic plaques, and among peripheric macrophages (with or without foam cells induction by oxidized low density lipoprotein, oxLDL) from healthy and atherosclerotic subjects. Two independent datasets, GSE28829 and GSE9874, were selected from gene expression omnibus (http://www.ncbi.nlm.nih.gov/geo/) repository. Functional interactions were obtained with STRING (http://string-db.org/) and Medusa (http://coot.embl.de/medusa/). Statistical analysis was performed with ViaComplex^® (http://lief.if.ufrgs.br/pub/biosoftwares/viacomplex/) and gene score enrichment analysis (http://www.broadinstitute.org/gsea/index.jsp). Bootstrap analysis demonstrated that the activity (expression) of HAG and M1 gene sets were significantly increased in advance compared to early atherosclerotic plaque. Increased expressions of HAG, M1, and M2 gene sets were found in peripheric macrophages from atherosclerotic subjects compared to peripheric macrophages from healthy subjects, while only M1 gene set was increased in foam cells from atherosclerotic subjects compared to foam cells from healthy subjects. However, M1 gene set was decreased in foam cells from healthy subjects compared to peripheric macrophages from healthy subjects, while no differences were found in foam cells from atherosclerotic subjects compared to peripheric macrophages from atherosclerotic subjects. Our data suggest that, different to cancer, in atherosclerosis there is no M1 or M2 polarization of macrophages. Actually, M1 and M2 phenotype are equally induced, what is an important aspect to better understand the disease progression, and can help to develop new therapeutic approaches.