Predicting gene regulation by sigma factors in Bacillus subtilis from genome-wide data.

MOTIVATION: Sigma factors regulate the expression of genes in Bacillus subtilis at the transcriptional level. We assess the accuracy of a fold-change analysis, Bayesian networks, dynamic models and supervised learning based on coregulation in predicting gene regulation by sigma factors from gene expression data. To improve the prediction accuracy, we combine sequence information with expression data by adding their log-likelihood scores and by using a logistic regression model. We use the resulting score function to discover currently unknown gene regulations by sigma factors. RESULTS: The coregulation-based supervised learning method gave the most accurate prediction of sigma factors from expression data. We found that the logistic regression model effectively combines expression data with sequence information. In a genome-wide search, highly significant logistic regression scores were found for several genes whose transcriptional regulation is currently unknown. We provide the corresponding RNA polymerase binding sites to enable a straightforward experimental verification of these predictions.

Induction of microRNAs, mir-155, mir-222, mir-424 and mir-503, promotes monocytic differentiation through combinatorial regulation.

Acute myeloid leukemia (AML) involves a block in terminal differentiation of the myeloid lineage and uncontrolled proliferation of a progenitor state. Using phorbol myristate acetate (PMA), it is possible to overcome this block in THP-1 cells (an M5-AML containing the MLL-MLLT3 fusion), resulting in differentiation to an adherent monocytic phenotype. As part of FANTOM4, we used microarrays to identify 23 microRNAs that are regulated by PMA. We identify four PMA-induced microRNAs (mir-155, mir-222, mir-424 and mir-503) that when overexpressed cause cell-cycle arrest and partial differentiation and when used in combination induce additional changes not seen by any individual microRNA. We further characterize these pro-differentiative microRNAs and show that mir-155 and mir-222 induce G2 arrest and apoptosis, respectively. We find mir-424 and mir-503 are derived from a polycistronic precursor mir-424-503 that is under repression by the MLL-MLLT3 leukemogenic fusion. Both of these microRNAs directly target cell-cycle regulators and induce G1 cell-cycle arrest when overexpressed in THP-1. We also find that the pro-differentiative mir-424 and mir-503 downregulate the anti-differentiative mir-9 by targeting a site in its primary transcript. Our study highlights the combinatorial effects of multiple microRNAs within cellular systems.

Bayesian joint prediction of associated transcription factors in Bacillus subtilis.

Sigma factors, often in conjunction with other transcription factors, regulate gene expression in prokaryotes at the transcriptional level. Specific transcription factors tend to co-occur with specific sigma factors. To predict new members of the transcription factor regulon, we applied Bayes rule to combine the Bayesian probability of sigma factor prediction calculated from microarray data and the sigma factor binding sequence motif, the motif score of the transcription factor associated with the sigma factor, the empirically determined distance between the transcription start site to the cis-regulatory region, and the tendency for specific sigma factors and transcription factors to co-occur. By combining these information sources, we improve the accuracy of predicting regulation by transcription factors, and also confirm the sigma factor prediction. We applied our proposed method to all genes in Bacillus subtilis to find currently unknown gene regulations by transcription factors and sigma factors.

Statistical analysis of a small set of time-ordered gene expression data using linear splines.

MOTIVATION: Recently, the temporal response of genes to changes in their environment has been investigated using cDNA microarray technology by measuring the gene expression levels at a small number of time points. Conventional techniques for time series analysis are not suitable for such a short series of time-ordered data. The analysis of gene expression data has therefore usually been limited to a fold-change analysis, instead of a systematic statistical approach. METHODS: We use the maximum likelihood method together with Akaike's Information Criterion to fit linear splines to a small set of time-ordered gene expression data in order to infer statistically meaningful information from the measurements. The significance of measured gene expression data is assessed using Student's t-test. RESULTS: Previous gene expression measurements of the cyanobacterium Synechocystis sp. PCC6803 were reanalyzed using linear splines. The temporal response was identified of many genes that had been missed by a fold-change analysis. Based on our statistical analysis, we found that about four gene expression measurements or more are needed at each time point.

Open source clustering software.

SUMMARY: We have implemented k-means clustering, hierarchical clustering and self-organizing maps in a single multipurpose open-source library of C routines, callable from other C and C++ programs. Using this library, we have created an improved version of Michael Eisen's well-known Cluster program for Windows, Mac OS X and Linux/Unix. In addition, we generated a Python and a Perl interface to the C Clustering Library, thereby combining the flexibility of a scripting language with the speed of C. AVAILABILITY: The C Clustering Library and the corresponding Python C extension module Pycluster were released under the Python License, while the Perl module Algorithm::Cluster was released under the Artistic License. The GUI code Cluster 3.0 for Windows, Macintosh and Linux/Unix, as well as the corresponding command-line program, were released under the same license as the original Cluster code. The complete source code is available at http://bonsai.ims.u-tokyo.ac.jp/mdehoon/software/cluster. Alternatively, Algorithm::Cluster can be downloaded from CPAN, while Pycluster is also available as part of the Biopython distribution.

Predicting the operon structure of Bacillus subtilis using operon length, intergene distance, and gene expression information.

We predict the operon structure of the Bacillus subtilis genome using the average operon length, the distance between genes in base pairs, and the similarity in gene expression measured in time course and gene disruptant experiments. By expressing the operon prediction for each method as a Bayesian probability, we are able to combine the four prediction methods into a Bayesian classifier in a statistically rigorous manner. The discriminant value for the Bayesian classifier can be chosen by considering the associated cost of misclassifying an operon or a non-operon gene pair. For equal costs, an overall accuracy of 88.7% was found in a leave-one-out analysis for the joint Bayesian classifier, whereas the individual information sources yielded accuracies of 58.1%, 83.1%, 77.3%, and 71.8% respectively. The predicted operon structure based on the joint Bayesian classifier is available from the DBTBS database (http://dbtbs.hgc.jp).