doi:10.1093/nar/gkl1019

* Corresponding authors

We present a kernel-based approach to the classification of a time series of gene expression profiles. Our method takes into account the dynamic evolution over time as well as the temporal characteristics of the data. More specifically, we model the evolution of the gene expression profiles as a Linear Time Invariant (LTI) dynamical system and estimate its model parameters. A kernel on dynamical systems is then used to classify these time series. We successfully test our approach on a published dataset to predict response to drug therapy in Multiple Sclerosis patients. For pharmacogenomics, our method offers a huge potential for advanced computational tools in disease diagnosis, and disease and drug therapy outcome prognosis. 1.

tools for microarray-based gene expression analysis

elegans data sources using compositional data mining

The Gene Expression Database (GXD) provides the scientific community with an extensive and easily searchable database of gene expression information about the mouse. Its primary emphasis is on developmental studies. By integrating different types of expression data, GXD aims to provide comprehensive information about expression patterns of transcripts and proteins in wild-type and mutant mice. Integration with the other Mouse Genome Informatics (MGI) databases places the gene expression information in the context of genetic, sequence, functional and phenotypic information, enabling valuable insights into the molecular biology that underlies developmental and disease processes. In recent years the utility of GXD has been greatly enhanced by a large increase in data content, obtained from the literature and provided by researchers doing large-scale in situ and cDNA screens. In addition, we have continued to refine our query and display features to make it easier for users to interrogate the data. GXD is available through the MGI web site at

High-throughput biological screens are yielding ever-growing streams of information about multiple aspects of cellular activity. As more and more categories of datasets come online, there is a corresponding multitude of ways in which inferences can be chained across them, motivating the need for compositional data mining algorithms. In this paper, we argue that such compositional data mining can be effectively realized by functionally cascading redescription mining and biclustering algorithms as primitives. Both these primitives mirror shifts of vocabulary that can be composed in arbitrary ways to create rich chains of inferences. Given a relational database and its schema, we show how the schema can be automatically compiled into a compositional data mining program, and how different domains in the schema can be related through logical sequences of biclustering and redescription invocations. This feature allows us to rapidly prototype new data mining applications, yielding greater understanding of scientific datasets. We describe two applications of compositional data mining: (i) matching terms across categories of the Gene Ontology and (ii) understanding the molecular mechanisms underlying stress response in human cells.

Abstract. Recent technological advances have enabled high-throughput measurements of protein-protein interactions in the cell, producing protein interaction networks for various species at an ever increasing pace. However, common technologies like yeast two-hybrid can experience high rates of false positive detection. To combat these errors, many methods have been developed which associate confidence scores with each interaction. Here we perform the first comparative analysis and performance assessment among these different methods using the fact that interacting proteins have similar biological attributes such as function, expression, and evolutionary conservation. We also introduce a new measure, the signal to noise ratio of protein complexes embedded in each network, to assess the quality of the different methods. We observe that utilizing any probability scheme is always more beneficial than assuming all observed interactions to be real. Also, schemes that assign probabilities to individual interactions generally perform better than those assessing the reliability of a set of interactions obtained from an experiment or a database. 1

The task of deriving a functional annotation for genes is complex as their involvement in various processes depends on multiple factors such as environmental conditions and genetic backup mechanisms. This study employs a large-scale model of the metabolism of Saccharomyces cerevisiae to investigate the function of yeast genes and derive a condition-dependent annotation (CDA) for their involvement in major metabolic processes under various genetic and environmental conditions. The resulting CDA is validated on a large scale and is shown to be superior to the corresponding Gene Ontology (GO) annotation, by showing that genes annotated with the same CDA term tend to be more coherently conserved in evolution and display greater expression coherency than those annotated with the same GO term. The CDA gives rise to new kinds of functional condition-dependent metabolic pathways, some of which are described and further examined via substrate auxotrophy measurements of knocked-out strains. The CDA presented is likely to serve as a new reference source for metabolic gene annotation. [Supplemental material is available online at www.genome.org.] In recent years, high-throughput techniques have provided a wealth of data on the expression and activity of genes and proteins. The task of inferring the involvement of gene products in various cellular processes, commonly referred to as functional

At some time during the course of any bioinformatics project, a researcher must go to a database that houses biological data. Whether it is a local database that records internal data from that laboratory’s experiments or a public database accessed through the Internet, such as NCBI’s GenBank (1) or EBI’s EMBL (2), researchers use biological databases for multiple reasons. One of the founding reasons for the fields of bioinformatics and computational biology was the need for management of biological data. In the past several decades, biological disciplines, including molecular biology and biochemistry, have generated massive amounts of data that are difficult to organize for efficient search, query, and analysis. If we trace the histories of both database development and the development of biochemical databases, we