CiteULike: oannes's library [44 articles]

A Gentle Tutorial on the EM Algorithm and its Application to Parameter Estimation for Gaussian Mixture and Hidden Markov Models

J Bilmes — 2005-10-22T03:31:57-00:00

(1997)

We describe the maximum-likelihood parameter estimation problem and how the ExpectationMaximization (EM) algorithm can be used for its solution. We first describe the abstract form of the EM algorithm as it is often given in the literature. We then develop the EM parameter estimation procedure for two applications: 1) finding the parameters of a mixture of Gaussian densities, and 2) finding the parameters of a hidden Markov model (HMM) (i.e., the Baum-Welch algorithm) for both discrete and...

Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing

Yoav Benjamini — 2007-01-15T14:09:52-00:00

The common approach to the multiplicity problem calls for controlling the familywise error rate (FWER). This approach, though, has faults, and we point out a few. A different approach to problems of multiple significance testing is presented. It calls for controlling the expected proportion of falsely rejected hypotheses-the false discovery rate. This error rate is equivalent to the FWER when all hypotheses are true but is smaller otherwise. Therefore, in problems where the control of the false discovery rate rather than that of the FWER is desired, there is potential for a gain in power. A simple sequential Bonferroni-type procedure is proved to control the false discovery rate for independent test statistics, and a simulation study shows that the gain in power is substantial. The use of the new procedure and the appropriateness of the criterion are illustrated with examples.

Statistical modeling of biomedical corpora: mining the Caenorhabditis Genetic Center Bibliography for genes related to life span.

D Blei — 2006-05-15T07:28:54-00:00

BMC Bioinformatics, Vol. 7, No. 1. (8 May 2006)

ABSTRACT: BACKGROUND: The statistical modeling of biomedical corpora could yield integrated, coarse-to-fine views of biological phenomena that complement discoveries made from analysis of molecular sequence and profiling data. Here, the potential of such modeling is demonstrated by examining the 5,225 free-text items in the Caenorhabditis Genetic Center (CGC) Bibliography using techniques from statistical information retrieval. Items in the CGC biomedical text corpus were modeled using the Latent Dirichlet Allocation (LDA) model. LDA is a hierarchical Bayesian model which represents a document as a random mixture over latent topics; each topic is characterized by a distribution over words. RESULTS: An LDA model estimated from CGC items had better predictive performance than two standard models (unigram and mixture of unigrams) trained using the same data. To illustrate the practical utility of LDA models of biomedical corpora, a trained CGC LDA model was used for a retrospective study of nematode genes known to be associated with life span modification. Corpus-, document-, and word-level LDA parameters were combined with terms from the Gene Ontology to enhance the explanatory value of the CGC LDA model, and to suggest additional candidates for age-related genes. A novel, pairwise document similarity measure based on the posterior distribution on the topic simplex was formulated and used to search the CGC database for "homologs" of a "query" document discussing the life span-modifying clk-2 gene. Inspection of these document homologs enabled and facilitated the production of hypotheses about the function and role of clk-2. CONCLUSIONS: Like other graphical models for genetic, genomic and other types of biological data, LDA provides a method for extracting unanticipated insights and generating predictions amenable to subsequent experimental validation.

Identifying biological concepts from a protein-related corpus with a probabilistic topic model.

B Zheng — 2006-06-20T21:22:40-00:00

BMC Bioinformatics, Vol. 7 (2006)

BACKGROUND: Biomedical literature, e.g., MEDLINE, contains a wealth of knowledge regarding functions of proteins. Major recurring biological concepts within such text corpora represent the domains of this body of knowledge. The goal of this research is to identify the major biological topics/concepts from a corpus of protein-related MEDLINE titles and abstracts by applying a probabilistic topic model. RESULTS: The latent Dirichlet allocation (LDA) model was applied to the corpus. Based on the Bayesian model selection, 300 major topics were extracted from the corpus. The majority of identified topics/concepts was found to be semantically coherent and most represented biological objects or concepts. The identified topics/concepts were further mapped to the controlled vocabulary of the Gene Ontology (GO) terms based on mutual information. CONCLUSION: The major and recurring biological concepts within a collection of MEDLINE documents can be extracted by the LDA model. The identified topics/concepts provide parsimonious and semantically-enriched representation of the texts in a semantic space with reduced dimensionality and can be used to index text.

Finding genomic ontology terms in text using evidence content.

FM Couto — 2005-12-20T16:34:59-00:00

BMC Bioinformatics, Vol. 6 Suppl 1 (2005)

BACKGROUND: The development of text mining systems that annotate biological entities with their properties using scientific literature is an important recent research topic. These systems need first to recognize the biological entities and properties in the text, and then decide which pairs represent valid annotations. METHODS: This document introduces a novel unsupervised method for recognizing biological properties in unstructured text, involving the evidence content of their names. RESULTS: This document shows the results obtained by the application of our method to BioCreative tasks 2.1 and 2.2, where it identified Gene Ontology annotations and their evidence in a set of articles. CONCLUSION: From the performance obtained in BioCreative, we concluded that an automatic annotation system can effectively use our method to identify biological properties in unstructured text.

Automatically annotating documents with normalized gene lists.

J Crim — 2006-05-22T15:27:54-00:00

BMC Bioinformatics, Vol. 6 Suppl 1 (2005)

BACKGROUND: Document gene normalization is the problem of creating a list of unique identifiers for genes that are mentioned within a document. Automating this process has many potential applications in both information extraction and database curation systems. Here we present two separate solutions to this problem. The first is primarily based on standard pattern matching and information extraction techniques. The second and more novel solution uses a statistical classifier to recognize valid gene matches from a list of known gene synonyms. RESULTS: We compare the results of the two systems, analyze their merits and argue that the classification based system is preferable for many reasons including performance, simplicity and robustness. Our best systems attain a balanced precision and recall in the range of 74%-92%, depending on the organism.

Finding scientific topics.

TL Griffiths — 2006-06-16T19:50:07-00:00

Proc Natl Acad Sci U S A, Vol. 101 Suppl 1 (6 April 2004), pp. 5228-5235.

A first step in identifying the content of a document is determining which topics that document addresses. We describe a generative model for documents, introduced by Blei, Ng, and Jordan [Blei, D. M., Ng, A. Y. & Jordan, M. I. (2003) J. Machine Learn. Res. 3, 993-1022], in which each document is generated by choosing a distribution over topics and then choosing each word in the document from a topic selected according to this distribution. We then present a Markov chain Monte Carlo algorithm for inference in this model. We use this algorithm to analyze abstracts from PNAS by using Bayesian model selection to establish the number of topics. We show that the extracted topics capture meaningful structure in the data, consistent with the class designations provided by the authors of the articles, and outline further applications of this analysis, including identifying "hot topics" by examining temporal dynamics and tagging abstracts to illustrate semantic content.

Mixed-membership models of scientific publications.

E Erosheva — 2006-06-16T19:49:08-00:00

Proc Natl Acad Sci U S A, Vol. 101 Suppl 1 (6 April 2004), pp. 5220-5227.

PNAS is one of world's most cited multidisciplinary scientific journals. The PNAS official classification structure of subjects is reflected in topic labels submitted by the authors of articles, largely related to traditionally established disciplines. These include broad field classifications into physical sciences, biological sciences, social sciences, and further subtopic classifications within the fields. Focusing on biological sciences, we explore an internal soft-classification structure of articles based only on semantic decompositions of abstracts and bibliographies and compare it with the formal discipline classifications. Our model assumes that there is a fixed number of internal categories, each characterized by multinomial distributions over words (in abstracts) and references (in bibliographies). Soft classification for each article is based on proportions of the article's content coming from each category. We discuss the appropriateness of the model for the PNAS database as well as other features of the data relevant to soft classification.

ITTACA: a new database for integrated tumor transcriptome array and clinical data analysis.

A Elfilali — 2006-06-06T20:04:40-00:00

Nucleic Acids Res, Vol. 34, No. Database issue. (1 January 2006)

Transcriptome microarrays have become one of the tools of choice for investigating the genes involved in tumorigenesis and tumor progression, as well as finding new biomarkers and gene expression signatures for the diagnosis and prognosis of cancer. Here, we describe a new database for Integrated Tumor Transcriptome Array and Clinical data Analysis (ITTACA). ITTACA centralizes public datasets containing both gene expression and clinical data. ITTACA currently focuses on the types of cancer that are of particular interest to research teams at Institut Curie: breast carcinoma, bladder carcinoma and uveal melanoma. A web interface allows users to carry out different class comparison analyses, including the comparison of expression distribution profiles, tests for differential expression and patient survival analyses. ITTACA is complementary to other databases, such as GEO and SMD, because it offers a better integration of clinical data and different functionalities. It also offers more options for class comparison analyses when compared with similar projects such as Oncomine. For example, users can define their own patient groups according to clinical data or gene expression levels. This added flexibility and the user-friendly web interface makes ITTACA especially useful for comparing personal results with the results in the existing literature. ITTACA is accessible online at http://bioinfo.curie.fr/ittaca.

Interplay between VHL/HIF1α and Wnt/β-catenin pathways during colorectal tumorigenesis

RH Giles — 2006-01-11T15:55:34-00:00

Oncogene, Vol. aop, No. current.

Inactivation of von Hippel-Lindau Gene Induces Constitutive Phosphorylation of MET Protein in Clear Cell Renal Carcinoma

Noboru Nakaigawa — 2006-06-06T19:20:12-00:00

Cancer Res, Vol. 66, No. 7. (1 April 2006), pp. 3699-3705.

It is well known that inactivation of von Hippel-Lindau (VHL) gene predisposes for human clear cell renal carcinoma (CCRC). However, details about critical roles of VHL inactivation during tumorigenesis are still unknown. MET protein is a tyrosine kinase receptor for hepatocyte growth factor/scatter factor (HGF/SF), which regulates cell growth, cell morphology, and cell motility. We showed that MET protein overexpressed in CCRC cells was phosphorylated without HGF/SF. This constitutive phosphorylation of MET protein in CCRC cells was inhibited by the rescue of exogenous wild-type VHL gene without a decrease in expression level of MET protein. Interestingly, wild-type VHL gene suppressed the phosphorylation of MET protein only under high cell density conditions. Additionally, MET protein activated by the inactivation of VHL gene modified cell adherence, including N-cadherin and beta-catenin. When activation of MET protein in CCRC cells was inhibited by the MET inhibitor K252a, the growth of CCRC cells in vitro and the tumorigenesis induced by CCRC cells in nude mice were suppressed. From these results, we concluded that inactivation of VHL gene induced constitutive phosphorylation of MET protein and modified intercellular adherence structure to trigger the cell growth released from contact inhibition, finally resulting in tumorigenesis. This is one of the mechanisms of CCRC oncogenesis, and MET protein has potential as a molecular target for novel CCRC therapies. (Cancer Res 2006; 66(7): 3699-705) 10.1158/0008-5472.CAN-05-0617

The transcriptional regulatory code of eukaryotic cells - insights from genome-wide analysis of chromatin organization and transcription factor binding

Leah Barrera — 2006-05-21T10:55:49-00:00

Current Opinion in Cell Biology, Vol. 18, No. 3. (June 2006), pp. 291-298.

Control of eukaryotic gene expression involves combinatorial interactions between transcription factors and regulatory sequences in the genome. In addition, chromatin structure and modification states play key roles in determining the competence of transcription. The term 'transcriptional regulatory code' has been used to describe the interplay of these events in the complex control of transcription. With the maturation of methods for detecting in vivo protein-DNA interactions on a genome-wide scale, detailed maps of chromatin features and transcription factor localization over entire genomes of eukaryotic cells are enriching our understanding of the properties and nature of this transcriptional regulatory code. The rapidly growing number of maps has revealed the dynamic nature of nucleosome composition and chromatin remodeling at regulatory regions and highlighted some unexpected properties of transcriptional regulatory networks in eukaryotic cells.

Composite Module Analyst: a fitness-based tool for identification of transcription factor binding site combinations.

A Kel — 2006-06-01T02:18:32-00:00

Bioinformatics, Vol. 22, No. 10. (15 May 2006), pp. 1190-1197.

MOTIVATION: Functionally related genes involved in the same molecular-genetic, biochemical or physiological process are often regulated coordinately. Such regulation is provided by precisely organized binding of a multiplicity of special proteins [transcription factors (TFs)] to their target sites (cis-elements) in regulatory regions of genes. Cis-element combinations provide a structural basis for the generation of unique patterns of gene expression. RESULTS: Here we present a new approach for defining promoter models based on the composition of TF binding sites and their pairs. We utilize a multicomponent fitness function for selection of the promoter model that fits best to the observed gene expression profile. We demonstrate examples of successful application of the fitness function with the help of a genetic algorithm for the analysis of functionally related or co-expressed genes as well as testing on simulated and permutated data. AVAILABILITY: The CMA program is freely available for non-commercial users. URL http://www.gene-regulation.com/pub/programs.html#CMAnalyst. It is also a part of the commercial system ExPlaintrade mark (www.biobase.de) designed for causal analysis of gene expression data. CONTACT: alexander.kel@biobase-international.com.

Human Transcription Factors Contain a High Fraction of Intrinsically Disordered Regions Essential for Transcriptional Regulation.

Yoshiaki Minezaki — 2006-06-01T02:12:22-00:00

J Mol Biol (25 April 2006)

Human transcriptional regulation factors, such as activators, repressors, and enhancer-binding factors are quite different from their prokaryotic counterparts in two respects: the average sequence in human is more than twice as long as that in prokaryotes, while the fraction of sequence aligned to domains of known structure is 31% in human transcription factors (TFs), less than half of that in bacterial TFs (72%). Intrinsically disordered (ID) regions were identified by a disorder-prediction program, and were found to be in good agreement with available experimental data. Analysis of 401 human TFs with experimental evidence from the Swiss-Prot database showed that as high as 49% of the entire sequence of human TFs is occupied by ID regions. More than half of the human TFs consist of a small DNA binding domain (DBD) and long ID regions frequently sandwiching unassigned regions. The remaining TFs have structural domains in addition to DBDs and ID regions. Experimental studies, particularly those with NMR, revealed that the transactivation domains in unbound TFs are usually unstructured, but become structured upon binding to their partners. The sequences of human and mouse TF orthologues are 90.5% identical despite a high incidence of ID regions, probably reflecting important functional roles played by ID regions. In general ID regions occupy a high fraction in TFs of eukaryotes, but not in prokaryotes. Implications of this dichotomy are discussed in connection with their functional roles in transcriptional regulation and evolution.

BioMed Central | Abstract | 1471-2105-7-229 | Meta-analysis discovery of tissue-specific DNA sequence motifs from mammalian gene expression data

2006-05-26T20:47:37-00:00

Transcription in four dimensions: nuclear receptor-directed initiation of gene expression.

R Métivier — 2006-05-25T16:26:07-00:00

EMBO Rep, Vol. 7, No. 2. (February 2006), pp. 161-167.

Regulated gene expression, achieved through the coordinated assembly of transcription factors, co-regulators and the basal transcription machinery on promoters, is an initial step in accomplishing cell specificity and homeostasis. Traditional models of transcriptional regulation tend to be static, although gene expression profiles change with time to adapt to developmental and environmental cues. Furthermore, biochemical and structural studies have determined that initiation of transcription progresses through a series of ordered events. By integrating time into the analysis of transcription, chromatin immunoprecipitation assays and live-cell imaging techniques have revealed the dynamic, cooperative, functionally redundant and cyclical nature of gene expression. In this review, we present a dynamic model of gene transcription that integrates data obtained by these two techniques.

SUMO and ubiquitin in the nucleus: different functions, similar mechanisms?

Grace Gill — 2006-05-25T16:12:34-00:00

Genes Dev., Vol. 18, No. 17. (1 September 2004), pp. 2046-2059.

The small ubiquitin-related modifier SUMO posttranslationally modifies many proteins with roles in diverse processes including regulation of transcription, chromatin structure, and DNA repair. Similar to nonproteolytic roles of ubiquitin, SUMO modification regulates protein localization and activity. Some proteins can be modified by SUMO and ubiquitin, but with distinct functional consequences. It is possible that the effects of ubiquitination and SUMOylation are both largely due to binding of proteins bearing specific interaction domains. Both modifications are reversible, and in some cases dynamic cycles of modification may be required for activity. Studies of SUMO and ubiquitin in the nucleus are yielding new insights into regulation of gene expression, genome maintenance, and signal transduction. 10.1101/gad.1214604

Contrasting effects on HIF-1alpha regulation by disease-causing pVHL mutations correlate with patterns of tumourigenesis in von Hippel-Lindau disease.

SC Clifford — 2006-05-23T19:34:46-00:00

Hum Mol Genet, Vol. 10, No. 10. (1 May 2001), pp. 1029-1038.

The von Hippel-Lindau tumour suppressor gene product (pVHL) associates with the elongin B and C and Cul2 proteins to form a ubiquitin-ligase complex (VCBC). To date, the only VCBC substrates identified are the hypoxia-inducible factor alpha subunits (HIF-1alpha and HIF-2alpha). However, pVHL is thought to have multiple functions and the significance of HIF-1alpha and HIF-2alpha regulation for tumour suppressor activity has not been defined. VHL disease is characterized by distinct clinical subtypes. Thus haemangioblastomas (HABs) and renal cell carcinoma (RCC) but not phaeochromocytoma (PHE) occur in type 1 VHL disease. Type 2 subtypes are characterized by PHE susceptibility but differ with respect to additional tumours (type 2A, PHE+HAB but not RCC; type 2B, PHE+ HAB+RCC; type 2C, PHE only). We investigated in detail the effect of 13 naturally occurring VHL mutations (11 missense), representing each phenotypic subclass, on HIF-alpha subunit regulation. Consistent effects on pVHL function were observed for all mutations within each subclass. Mutations associated with the PHE-only phenotype (type 2C) promoted HIF-alpha ubiquitylation in vitro and demonstrated wild-type binding patterns with pVHL interacting proteins, suggesting that loss of other pVHL functions are necessary for PHE susceptibility. Mutations causing HAB susceptibility (types 1, 2A and 2B) demonstrated variable effects on HIF-alpha subunit and elongin binding, but all resulted in defective HIF-alpha regulation and loss of p220 (fibronectin) binding. All RCC-associated mutations caused complete HIF-alpha dysregulation and loss of p220 (fibronectin) binding. Our findings are consistent with impaired ability to degrade HIF-alpha subunit being required for HAB development and RCC susceptibility.

p53 Stabilization and Transactivation by a von Hippel-Lindau Protein.

JS Roe — 2006-05-23T19:31:11-00:00

Mol Cell, Vol. 22, No. 3. (5 May 2006), pp. 395-405.

von Hippel-Lindau (VHL) disease is a rare autosomal dominant cancer syndrome. Although hypoxia-inducible factor-alpha (HIFalpha) is a well-documented substrate of von Hippel-Lindau tumor suppressor protein (pVHL), it remains unclear whether the dysregulation of HIF is sufficient to account for de novo tumorigenesis in VHL-deleted cells. Here we found that pVHL directly associates with and stabilizes p53 by suppressing Mdm2-mediated ubiquitination and nuclear export of p53. Moreover, upon genotoxic stress, pVHL invoked an interaction between p53 and p300 and the acetylation of p53, which ultimately led to an increase in p53 transcriptional activity and p53-mediated cell cycle arrest and apoptosis. These results suggest that the tumor suppressor pVHL has an unexpected function to upregulate the tumor suppressor p53.

Hypoxia-inducible factors: where, when and why?

JM Gleadle — 2006-05-23T19:14:29-00:00

Vol. 69, No. 1. (0000), pp. 15-17.

Activation of TRAP/Mediator Subunit TRAP220/Med1 Is Regulated by Mitogen-Activated Protein Kinase-Dependent Phosphorylation

Pradeep Pandey — 2006-05-18T22:15:32-00:00

Mol. Cell. Biol., Vol. 25, No. 24. (15 December 2005), pp. 10695-10710.

The TRAP/Mediator coactivator complex serves as a molecular bridge between gene-specific activators and RNA polymerase II. TRAP220/Med1 is a key component of TRAP/Mediator that targets the complex to nuclear hormone receptors and other types of activators. We show here that human TRAP220/Med1 is a specific substrate for extracellular signal-regulated kinase (ERK) of the mitogen-activated protein kinase (MAPK) family. We demonstrate that ERK phosphorylates TRAP220/Med1 in vivo at two specific sites: threonine 1032 and threonine 1457. Importantly, we found that ERK phosphorylation significantly increases the stability and half-life of TRAP220/Med1 in vivo and correlates with increased thyroid hormone receptor-dependent transcription. Furthermore, ERK phosphorylates TRAP220/Med1 in a cell cycle-dependent manner, resulting in peak levels of expression during the G2/M phase of the cell cycle. ERK phosphorylation of ectopic TRAP220/Med1 also triggered shuttling into the nucleolus, thus suggesting that ERK may regulate TRAP220/Med1 subnuclear localization. Finally, we observed that ERK phosphorylation of TRAP220/Med1 stimulates its intrinsic transcriptional coactivation activity. We propose that ERK-mediated phosphorylation is a regulatory mechanism that controls TRAP220/Med1 expression levels and modulates its functional activity.

Quantitative evaluation of protein-DNA interactions using an optimized knowledge-based potential.

Z Liu — 2006-05-18T04:57:23-00:00

Nucleic Acids Res, Vol. 33, No. 2. (2005), pp. 546-558.

Computational evaluation of protein-DNA interaction is important for the identification of DNA-binding sites and genome annotation. It could validate the predicted binding motifs by sequence-based approaches through the calculation of the binding affinity between a protein and DNA. Such an evaluation should take into account structural information to deal with the complicated effects from DNA structural deformation, distance-dependent multi-body interactions and solvation contributions. In this paper, we present a knowledge-based potential built on interactions between protein residues and DNA tri-nucleotides. The potential, which explicitly considers the distance-dependent two-body, three-body and four-body interactions between protein residues and DNA nucleotides, has been optimized in terms of a Z-score. We have applied this knowledge-based potential to evaluate the binding affinities of zinc-finger protein-DNA complexes. The predicted binding affinities are in good agreement with the experimental data (with a correlation coefficient of 0.950). On a larger test set containing 48 protein-DNA complexes with known experimental binding free energies, our potential has achieved a high correlation coefficient of 0.800, when compared with the experimental data. We have also used this potential to identify binding motifs in DNA sequences of transcription factors (TF). The TFs in 79.4% of the known TF-DNA complexes have accurately found their native binding sequences from a large pool of DNA sequences. When tested in a genome-scale search for TF-binding motifs of the cyclic AMP regulatory protein (CRP) of Escherichia coli, this potential ranks all known binding motifs of CRP in the top 15% of all candidate sequences.

Modeling gene expression control using Omes Law

Harmen Bussemaker — 2006-04-25T15:45:52-00:00

Mol Syst Biol, Vol. 2, No. 1. (11 April 2006), pp. msb4100055-E1-msb4100055-E3.

Local Regulatory Variation in Saccharomyces cerevisiae.

James Ronald — 2005-10-07T21:11:53-00:00

PLoS Genet, Vol. 1, No. 2. (19 August 2005)

Naturally occurring sequence variation that affects gene expression is an important source of phenotypic differences among individuals within a species. We and others have previously shown that such regulatory variation can occur both at the same locus as the gene whose expression it affects (local regulatory variation) and elsewhere in the genome at trans-acting factors. Here we present a detailed analysis of genome-wide local regulatory variation in Saccharomyces cerevisiae. We used genetic linkage analysis to show that nearly a quarter of all yeast genes contain local regulatory variation between two divergent strains. We measured allele-specific expression in a diploid hybrid of the two strains for 77 genes showing strong self-linkage and found that in 52%-78% of these genes, local regulatory variation acts directly in cis. We also experimentally confirmed one example in which local regulatory variation in the gene AMN1 acts in trans through a feedback loop. Genome-wide sequence analysis revealed that genes subject to local regulatory variation show increased polymorphism in the promoter regions, and that some but not all of this increase is due to polymorphisms in predicted transcription factor binding sites. Increased polymorphism was also found in the 3' untranslated regions of these genes. These findings point to the importance of cis-acting variation, but also suggest that there is a diverse set of mechanisms through which local variation can affect gene expression levels.

Trans-acting regulatory variation in Saccharomyces cerevisiae and the role of transcription factors

Gael Yvert — 2006-05-09T19:48:36-00:00

Nat Genet, Vol. 35, No. 1. (2003), pp. 57-64.

Genetic Dissection of Transcriptional Regulation in Budding Yeast

Rachel Brem — 2006-05-09T19:40:57-00:00

Science, Vol. 296, No. 5568. (26 April 2002), pp. 752-755.

10.1126/science.1069516

Learning the parts of objects by non-negative matrix factorization.

DD Lee — 2005-01-25T22:33:12-00:00

Nature, Vol. 401, No. 6755. (21 October 1999), pp. 788-791.

Is perception of the whole based on perception of its parts? There is psychological and physiological evidence for parts-based representations in the brain, and certain computational theories of object recognition rely on such representations. But little is known about how brains or computers might learn the parts of objects. Here we demonstrate an algorithm for non-negative matrix factorization that is able to learn parts of faces and semantic features of text. This is in contrast to other methods, such as principal components analysis and vector quantization, that learn holistic, not parts-based, representations. Non-negative matrix factorization is distinguished from the other methods by its use of non-negativity constraints. These constraints lead to a parts-based representation because they allow only additive, not subtractive, combinations. When non-negative matrix factorization is implemented as a neural network, parts-based representations emerge by virtue of two properties: the firing rates of neurons are never negative and synaptic strengths do not change sign.

Discovering semantic features in the literature: a foundation for building functional associations.

Monica Chagoyen — 2006-01-29T14:32:39-00:00

BMC Bioinformatics, Vol. 7, No. 1. (26 January 2006)

ABSTRACT: BACKGROUND: Experimental techniques such as DNA microarray, serial analysis of gene expression (SAGE) and mass spectrometry proteomics, among others, are generating large amounts of data related to genes and proteins at different levels. As in any other experimental approach, it is necessary to analyze these data in the context of previously known information about the biological entities under study. The literature is a particularly valuable source of information for experiment validation and interpretation. Therefore, the development of automated text mining tools to assist in such interpretation is one of the main challenges in current bioinformatics research. RESULTS: We present a method to create literature profiles for large sets of genes or proteins based on common semantic features extracted from a corpus of relevant documents. These profiles can be used to establish pair-wise similarities among genes, utilized in gene/protein classification or can be even combined with experimental measurements. Semantic features can be used by researchers to facilitate the understanding of the commonalities indicated by experimental results. Our approach is based on non-negative matrix factorization (NMF), a machine-learning algorithm for data analysis, capable of identifying local patterns that characterize a subset of the data. The literature is thus used to establish putative relationships among subsets of genes or proteins and to provide coherent justification for this clustering into subsets. We demonstrate the utility of the method by applying it to two independent and vastly different sets of genes. CONCLUSIONS: The presented method can create literature profiles from documents relevant to sets of genes. The representation of genes as additive linear combinations of semantic features allows for the exploration of functional associations as well as for clustering, suggesting a valuable methodology for the validation and interpretation of high-throughput experimental data.

Genome-wide computational prediction of transcriptional regulatory modules reveals new insights into human gene expression.

Mathieu Blanchette — 2006-04-27T19:23:55-00:00

Genome Res (10 April 2006)

The identification of regulatory regions is one of the most important and challenging problems toward the functional annotation of the human genome. In higher eukaryotes, transcription-factor (TF) binding sites are often organized in clusters called cis-regulatory modules (CRM). While the prediction of individual TF-binding sites is a notoriously difficult problem, CRM prediction has proven to be somewhat more reliable. Starting from a set of predicted binding sites for more than 200 TF families documented in Transfac, we describe an algorithm relying on the principle that CRMs generally contain several phylogenetically conserved binding sites for a few different TFs. The method allows the prediction of more than 118,000 CRMs within the human genome. A subset of these is shown to be bound in vivo by TFs using ChIP-chip. Their analysis reveals, among other things, that CRM density varies widely across the genome, with CRM-rich regions often being located near genes encoding transcription factors involved in development. Predicted CRMs show a surprising enrichment near the 3' end of genes and in regions far from genes. We document the tendency for certain TFs to bind modules located in specific regions with respect to their target genes and identify TFs likely to be involved in tissue-specific regulation. The set of predicted CRMs, which is made available as a public database called PReMod (http://genomequebec.mcgill.ca/PReMod), will help analyze regulatory mechanisms in specific biological systems.

Protein-DNA binding specificity predictions with structural models.

AV Morozov — 2006-04-27T19:19:28-00:00

Nucleic Acids Res, Vol. 33, No. 18. (2005), pp. 5781-5798.

Protein-DNA interactions play a central role in transcriptional regulation and other biological processes. Investigating the mechanism of binding affinity and specificity in protein-DNA complexes is thus an important goal. Here we develop a simple physical energy function, which uses electrostatics, solvation, hydrogen bonds and atom-packing terms to model direct readout and sequence-specific DNA conformational energy to model indirect readout of DNA sequence by the bound protein. The predictive capability of the model is tested against another model based only on the knowledge of the consensus sequence and the number of contacts between amino acids and DNA bases. Both models are used to carry out predictions of protein-DNA binding affinities which are then compared with experimental measurements. The nearly additive nature of protein-DNA interaction energies in our model allows us to construct position-specific weight matrices by computing base pair probabilities independently for each position in the binding site. Our approach is less data intensive than knowledge-based models of protein-DNA interactions, and is not limited to any specific family of transcription factors. However, native structures of protein-DNA complexes or their close homologs are required as input to the model. Use of homology modeling can significantly increase the extent of our approach, making it a useful tool for studying regulatory pathways in many organisms and cell types.

Annotating significant pairs of transcription factor binding sites in regulatory DNA.

K Rateitschak — 2006-04-27T19:17:17-00:00

In Silico Biol, Vol. 4, No. 4. (2004), pp. 479-487.

In the presented work we search for transcription factor binding sites (BS) by including additional information about typical BS patterns. The new proposed score combines the ordinary profile score based on TRANSFAC-matrices together with a score based on pairs of BS. The latter score positively weights pairs of BS that tend to occur together in many regulatory DNA-sequences, in contrast to a random background model. The empirical BS pair frequencies result from our evaluation of a large dataset of orthologous genes.

Distance preferences in the arrangement of binding motifs and hierarchical levels in organization of transcription regulatory information.

VJ Makeev — 2006-04-27T19:16:12-00:00

Nucleic Acids Res, Vol. 31, No. 20. (15 October 2003), pp. 6016-6026.

We explored distance preferences in the arrangement of binding motifs for five transcription factors (Bicoid, Krüppel, Hunchback, Knirps and Caudal) in a large set of Drosophila cis-regulatory modules (CRMs). Analysis of non-overlapping binding motifs revealed the presence of periodic signals specific to particular combinations of binding motifs. The most striking periodic signals (10 bp for Bicoid and 11 bp for Hunchback) suggest preferential positioning of some binding site combinations on the same side of the DNA helix. We also analyzed distance preferences in arrangements of highly correlated overlapping binding motifs, such as Bicoid and Krüppel. Based on the distance analysis, we extracted preferential binding site arrangements and proposed models for potential composite elements (CEs) and antagonistic motif pairs involved in the function of developmental CRMs. Our results suggest that there are distinct hierarchical levels in the organization of transcription regulatory information. We discuss the role of the hierarchy in understanding transcriptional regulation and in detection of transcription regulatory regions in genomes.

Predicting transcription factor synergism.

S Hannenhalli — 2006-04-27T19:13:09-00:00

Nucleic Acids Res, Vol. 30, No. 19. (1 October 2002), pp. 4278-4284.

Transcriptional regulation is mediated by a battery of transcription factor (TF) proteins, that form complexes involving protein-protein and protein-DNA interactions. Individual TFs bind to their cognate cis-elements or transcription factor-binding sites (TFBS). TFBS are organized on the DNA proximal to the gene in groups confined to a few hundred base pair regions. These groups are referred to as modules. Various modules work together to provide the combinatorial regulation of gene transcription in response to various developmental and environmental conditions. The sets of modules constitute a promoter model. Determining the TFs that preferentially work in concert as part of a module is an essential component of understanding transcriptional regulation. The TFs that act synergistically in such a fashion are likely to have their cis-elements co-localized on the genome at specific distances apart. We exploit this notion to predict TF pairs that are likely to be part of a transcriptional module on the human genome sequence. The computational method is validated statistically, using known interacting pairs extracted from the literature. There are 251 TFBS pairs up to 50 bp apart and 70 TFBS pairs up to 200 bp apart that score higher than any of the known synergistic pairs. Further investigation of 50 pairs randomly selected from each of these two sets using PubMed queries provided additional supporting evidence from the existing biological literature suggesting TF synergism for these novel pairs.

Annotating significant pairs of transcription factor binding sites in regulatory DNA

2006-04-27T19:10:02-00:00

Genome-wide prediction and analysis of function-specific transcription factor binding sites

2006-04-27T19:08:12-00:00

STEM: Short Time-series Expression Miner

2006-04-27T08:04:09-00:00

Statistic of Financial Markets

2006-04-27T08:00:15-00:00

Novel technique for preprocessing high dimensional time-course data from DNA microarray: mathematical model-based clustering.

K Hakamada — 2006-04-27T07:11:34-00:00

Bioinformatics, Vol. 22, No. 7. (1 April 2006), pp. 843-848.

MOTIVATION: Classifying genes into clusters depending on their expression profiles is one of the most important analysis techniques for microarray data. Because temporal gene expression profiles are indicative of the dynamic functional properties of genes, the application of clustering analysis to time-course data allows the more precise division of genes into functional classes. Conventional clustering methods treat the sampling data at each time point as data obtained under different experimental conditions without considering the continuity of time-course data between time periods t and t+1. Here, we propose a method designated mathematical model-based clustering (MMBC). RESULTS: The proposed method, designated MMBC, was applied to artificial data and time-course data obtained using Saccharomyces cerevisiae. Our method is able to divide data into clusters more accurately and coherently than conventional clustering methods. Furthermore, MMBC is more tolerant to noise than conventional clustering methods. AVAILABILITY: Software is available upon request. CONTACT: taizo@brs.kyushu-u.ac.jp.

R: Analysis of Multiple Time Course Data

2006-04-27T06:58:47-00:00

Differential and trajectory methods for time course gene expression data

Yulan Liang — 2005-06-21T10:29:26-00:00

Bioinformatics, Vol. 21, No. 13. (1 July 2005), pp. 3009-3016.

Integrated analysis of gene expression by Association Rules Discovery.

P Carmona-Saez — 2006-04-27T04:35:51-00:00

BMC Bioinformatics, Vol. 7 (2006)

BACKGROUND: Microarray technology is generating huge amounts of data about the expression level of thousands of genes, or even whole genomes, across different experimental conditions. To extract biological knowledge, and to fully understand such datasets, it is essential to include external biological information about genes and gene products to the analysis of expression data. However, most of the current approaches to analyze microarray datasets are mainly focused on the analysis of experimental data, and external biological information is incorporated as a posterior process. RESULTS: In this study we present a method for the integrative analysis of microarray data based on the Association Rules Discovery data mining technique. The approach integrates gene annotations and expression data to discover intrinsic associations among both data sources based on co-occurrence patterns. We applied the proposed methodology to the analysis of gene expression datasets in which genes were annotated with metabolic pathways, transcriptional regulators and Gene Ontology categories. Automatically extracted associations revealed significant relationships among these gene attributes and expression patterns, where many of them are clearly supported by recently reported work. CONCLUSION: The integration of external biological information and gene expression data can provide insights about the biological processes associated to gene expression programs. In this paper we show that the proposed methodology is able to integrate multiple gene annotations and expression data in the same analytic framework and extract meaningful associations among heterogeneous sources of data. An implementation of the method is included in the Engene software package.

A multi-step approach to time series analysis and gene expression clustering.

R Amato — 2006-04-27T04:18:18-00:00

Bioinformatics, Vol. 22, No. 5. (1 March 2006), pp. 589-596.

MOTIVATION: The huge growth in gene expression data calls for the implementation of automatic tools for data processing and interpretation. RESULTS: We present a new and comprehensive machine learning data mining framework consisting in a non-linear PCA neural network for feature extraction, and probabilistic principal surfaces combined with an agglomerative approach based on Negentropy aimed at clustering gene microarray data. The method, which provides a user-friendly visualization interface, can work on noisy data with missing points and represents an automatic procedure to get, with no a priori assumptions, the number of clusters present in the data. Cell-cycle dataset and a detailed analysis confirm the biological nature of the most significant clusters. AVAILABILITY: The software described here is a subpackage part of the ASTRONEURAL package and is available upon request from the corresponding author. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Schema for TFBS Conserved - HMR Conserved Transcription Factor Binding Sites

2006-04-24T17:43:45-00:00

Genome-wide prediction and characterization of interactions between transcription factors in Saccharomyces cerevisiae.

X Yu — 2006-02-24T15:58:20-00:00

Nucleic Acids Res, Vol. 34, No. 3. (2006), pp. 917-927.

Combinatorial regulation by transcription factor complexes is an important feature of eukaryotic gene regulation. Here, we propose a new method for identification of interactions between transcription factors (TFs) that relies on the relationship of their binding sites, and we test it using Saccharomyces cerevisiae as a model system. The algorithm predicts interacting TF pairs based on the co-occurrence of their binding motifs and the distance between the motifs in promoter sequences. This allows investigation of interactions between TFs without known binding motifs or expression data. With this approach, 300 significant interactions involving 77 TFs were identified. These included more than 70% of the known protein-protein interactions. Approximately half of the detected interacting motif pairs showed strong preferences for particular distances and orientations in the promoter sequences. These one dimensional features may reflect constraints on allowable spatial arrangements for protein-protein interactions. Evidence for biological relevance of the observed characteristic distances is provided by the finding that target genes with the same characteristic distances show significantly higher co-expression than those without preferred distances. Furthermore, the observed interactions were dynamic: most of the TF pairs were not constitutively active, but rather showed variable activity depending on the physiological condition of the cells. Interestingly, some TF pairs active in multiple conditions showed preferences for different distances and orientations depending on the condition. Our prediction and characterization of TF interactions may help to understand the transcriptional regulatory networks in eukaryotic systems.