New experimental and computational approaches to the analysis of gene expression.

Public and private EST (Expressed Sequence Tag) programs provide access to a large number of ESTs from a number of plant species, including Arabidopsis, corn, soybean, rice, wheat. In addition to the homology of each EST to genes in GenBank, information about homology to all other ESTs in the data base can be obtained. To estimate expression levels of genes represented in the DuPont EST data base we count the number of times each gene has been seen in different cDNA libraries, from different tissues, developmental stages or induction conditions. This quantitation of message levels is quite accurate for highly expressed messages and, unlike conventional Northern blots, allows comparison of expression levels between different genes. Lists of most highly expresses genes in different libraries can be compiled. Also, if EST data is available for cDNA libraries derived from different developmental stages, gene expression profiles across development can be assembled. We present an example of such a profile for soybean seed development. Gene expression data obtained from Electronic Northern analysis can be confirmed and extended beyond the realm of highly expressed genes by using high density DNA arrays. The ESTs identified as interesting can be arrayed on nylon or glass and probed with total labeled cDNA first strand from the tissue of interest. Two-color fluorescent labeling allows accurate mRNA ratio measurements. We are currently using the DNA array technology to study chemical induction of gene expression and the biosynthesis of oil, carbohydrate and protein in developing seeds.     

Identifying time-lagged gene clusters using gene expression data

MOTIVATION: Analysis of gene expression data can provide insights into the time-lagged co-regulation of genes/gene clusters. However, existing methods such as the Event Method and the Edge Detection Method are inefficient as they compare only two genes at a time. More importantly, they neglect some important information due to their scoring criterian. In this paper, we propose an efficient algorithm to identify time-lagged co-regulated gene clusters. The algorithm facilitates localized comparison and processes several genes simultaneously to generate detailed and complete time-lagged information for genes/gene clusters. RESULTS: We experimented with the time-series Yeast gene dataset and compared our algorithm with the Event Method. Our results show that our algorithm is not only efficient, but also delivers more reliable and detailed information on time-lagged co-regulation between genes/gene clusters. AVAILABILITY: The software is available upon request. CONTACT: jiliping@comp.nus.edu.sg SUPPLEMENTARY INFORMATION: Supplementary tables and figures for this paper can be found at http://www.comp.nus.edu.sg/~jiliping/p2.htm.     

Assessing reliability of gene clusters from gene expression data

The rapid development of microarray technologies has raised many challenging problems in experiment design and data analysis. Although many numerical algorithms have been successfully applied to analyze gene expression data, the effects of variations and uncertainties in measured gene expression levels across samples and experiments have been largely ignored in the literature. In this article, in the context of hierarchical clustering algorithms, we introduce a statistical resampling method to assess the reliability of gene clusters identified from any hierarchical clustering method. Using the clustering trees constructed from the resampled data, we can evaluate the confidence value for each node in the observed clustering tree. A majority-rule consensus tree can be obtained, showing clusters that only occur in a majority of the resampled trees. We illustrate our proposed methods with applications to two published data sets. Although the methods are discussed in the context of hierarchical clustering methods, they can be applied with other cluster-identification methods for gene expression data to assess the reliability of any gene cluster of interest. Electronic Publication     

A computational approach to measuring coherence of gene expression in pathways

This study uses a computational approach to analyze coherence of expression of genes in pathways. Microarray data were analyzed with respect to coherent gene expression in a group of genes defined as a pathway in the Kyoto Encyclopedia of Genes and Genomes (KEGG) database. Our hypothesis is that genes in the same pathway are more likely to be coordinately regulated than a randomly selected gene set. A correlation coefficient for each pair of genes in a pathway was estimated based on gene expression in normal or tumor samples, and statistically significant correlation coefficients were identified. The coherence indicator was defined as the ratio of the number of gene pairs in the pathway whose correlation coefficients are significant, divided by the total number of gene pairs in the pathway. We defined all genes that appeared in the KEGG pathways as a reference gene set. Our analysis indicated that the mean coherence indicator of pathways is significantly larger than the mean coherence indicator of random gene sets drawn from the reference gene set. Thus, the result supports our hypothesis. The significance of each individual pathway of n genes was evaluated by comparing its coherence indicator with coherence indicators of 1000 random permutation sets of n genes chosen from the reference gene set. We analyzed three data sets: two Affymetrix microarrays and one cDNA microarray. For each of the three data sets, statistically significant pathways were identified among all KEGG pathways. Seven of 96 pathways had a significant coherence indicator in normal tissue and 14 of 96 pathways had a significant coherence indicator in tumor tissue in all three data sets. The increase in the number of pathways with significant coherence indicators may reflect the fact that tumor cells have a higher rate of metabolism than normal cells. Five pathways involved in oxidative phosphorylation, ATP synthesis, protein synthesis, or RNA synthesis were coherent in both normal and tumor tissue, demonstrating that these are essential genes, a high level of expression of which is required regardless of cell type.     

Allelic variation in gene expression identified through computational analysis of the dbEST database

Differential expression between the two alleles of an individual and between people with different genotypes has been commonly observed. Quantitative differences in gene expression between people may provide the genetic basis for the phenotypic difference between individuals and may be the primary cause of complex diseases. In this paper, we developed a computational method to identify genes that displayed allelic variation in gene expression in human EST libraries. To model allele-specific gene expression, we first identified EST libraries in which both A and B alleles were expressed and then identified allelic variation in gene expression based on the EST counts for each allele using a binomial test. Among 1107 SNPs that had a sufficient number of ESTs for the analysis, 524 (47%) displayed allelic variation in at least one cDNA library. We verified experimentally the allelic variation in gene expression for 6 of these SNPs. The frequency of allelic variation observed in EST libraries was similar to the previous studies using the SNP chip and primer extension method. We found that genes that displayed allelic variation were distributed throughout the human genome and were enriched in certain chromosome regions. The SNPs and genes identified in this study will provide a rich source for evaluating the effects of those SNPs and associated haplotypes in human health and diseases.     

Heterologous expression of novobiocin and clorobiocin biosynthetic gene clusters

A method was developed for the heterologous expression of biosynthetic gene clusters in different Streptomyces strains and for the modification of these clusters by single or multiple gene replacements or gene deletions with unprecedented speed and versatility. Lambda-Red-mediated homologous recombination was used for genetic modification of the gene clusters, and the attachment site and integrase of phage phiC31 were employed for the integration of these clusters into the heterologous hosts. This method was used to express the gene clusters of the aminocoumarin antibiotics novobiocin and clorobiocin in the well-studied strains Streptomyces coelicolor and Streptomyces lividans, which, in contrast to the natural producers, can be easily genetically manipulated. S. coelicolor M512 derivatives produced the respective antibiotic in yields comparable to those of natural producer strains, whereas S. lividans TK24 derivatives were at least five times less productive. This method could also be used to carry out functional investigations. Shortening of the cosmids' inserts showed which genes are essential for antibiotic production.     

Mining gene expression data for positive and negative co-regulated gene clusters

MOTIVATION: Analysis of gene expression data can provide insights into the positive and negative co-regulation of genes. However, existing methods such as association rule mining are computationally expensive and the quality and quantities of the rules are sensitive to the support and confidence values. In this paper, we introduce the concept of positive and negative co-regulated gene cluster (PNCGC) that more accurately reflects the co-regulation of genes, and propose an efficient algorithm to extract PNCGCs. RESULTS: We experimented with the Yeast dataset and compared our resulting PNCGCs with the association rules generated by the Apriori mining algorithm. Our results show that our PNCGCs identify some missing co-regulations of association rules, and our algorithm greatly reduces the large number of rules involving uncorrelated genes generated by the Apriori scheme. AVAILABILITY: The software is available upon request.     

Robust cluster analysis of microarray gene expression data with the number of clusters determined biologically

MOTIVATION: The success of each method of cluster analysis depends on how well its underlying model describes the patterns of expression. Outlier-resistant and distribution-insensitive clustering of genes are robust against violations of model assumptions. RESULTS: A measure of dissimilarity that combines advantages of the Euclidean distance and the correlation coefficient is introduced. The measure can be made robust using a rank order correlation coefficient. A robust graphical method of summarizing the results of cluster analysis and a biological method of determining the number of clusters are also presented. These methods are applied to a public data set, showing that rank-based methods perform better than log-based methods. AVAILABILITY: Software is available from http://www.davidbickel.com.     

Constrained clusters of gene expression profiles with pathological features

MOTIVATION: Gene expression profiles should be useful in distinguishing variations in disease, since they reflect accurately the status of cells. The primary clustering of gene expression reveals the genotypes that are responsible for the proximity of members within each cluster, while further clustering elucidates the pathological features of the individual members of each cluster. However, since the first clustering process and the second classification step, in which the features are associated with clusters, are performed independently, the initial set of clusters may omit genes that are associated with pathologically meaningful features. Therefore, it is important to devise a way of identifying gene expression clusters that are associated with pathological features. RESULTS: We present the novel technique of 'itemset constrained clustering' (IC-Clustering), which computes the optimal cluster that maximizes the interclass variance of gene expression between groups, which are divided according to the restriction that only divisions that can be expressed using common features are allowed. This constraint automatically labels each cluster with a set of pathological features which characterize that cluster. When applied to liver cancer datasets, IC-Clustering revealed informative gene expression clusters, which could be annotated with various pathological features, such as 'tumor' and 'man', or 'except tumor' and 'normal liver function'. In contrast, the k-means method overlooked these clusters.     

From protoplasts to gene clusters

Advances in protoplast technology underpinned many crucial developments in our understanding of the molecular biology of filamentous fungi. This review follows one of these developments, namely the discovery and analysis of difuran toxin gene clusters. Our understanding of the biosynthetic pathway of the agriculturally important toxin, aflatoxin, has been dramatically enhanced by the use of protoplasts and protoplast-based gene transformation methods. Since the identification of the first pathway genes by complementation of mutants with transforming DNA, transformation has continued to play a critical role in the elucidation of gene function and regulation. But despite the wealth of knowledge accumulated so far some fundamental questions remain to be answered. How did these gene clusters evolve? What is the biological role of aflatoxin? The discovery of homologues of aflatoxin genes in other fungal species such as the pine needle pathogen Dothistroma septosporum may help to shed some light on these questions.     

How to write consistently boring scientific literature

Although scientists typically insist that their research is very exciting and adventurous when they talk to laymen and prospective students, the allure of this enthusiasm is too often lost in the predictable, stilted structure and language of their scientific publications. I present here, a top-10 list of recommendations for how to write consistently boring scientific publications. I then discuss why we should and how we could make these contributions more accessible and exciting.