SQUAT: A web tool to mine human, murine and avian SAGE data

Novel gene sets improve set-level classification of prokaryotic gene expression data

Background

Set-level classification of gene expression data has received significant attention recently. In this setting, high-dimensional vectors of features corresponding to genes are converted into lower-dimensional vectors of features corresponding to biologically interpretable gene sets. The dimensionality reduction brings the promise of a decreased risk of overfitting, potentially resulting in improved accuracy of the learned classifiers. However, recent empirical research has not confirmed this expectation. Here we hypothesize that the reported unfavorable classification results in the set-level framework were due to the adoption of unsuitable gene sets defined typically on the basis of the Gene ontology and the KEGG database of metabolic networks. We explore an alternative approach to defining gene sets, based on regulatory interactions, which we expect to collect genes with more correlated expression. We hypothesize that such more correlated gene sets will enable to learn more accurate classifiers.

Methods

We define two families of gene sets using information on regulatory interactions, and evaluate them on phenotype-classification tasks using public prokaryotic gene expression data sets. From each of the two gene-set families, we first select the best-performing subtype. The two selected subtypes are then evaluated on independent (testing) data sets against state-of-the-art gene sets and against the conventional gene-level approach.

Results

The novel gene sets are indeed more correlated than the conventional ones, and lead to significantly more accurate classifiers. The novel gene sets are indeed more correlated than the conventional ones, and lead to significantly more accurate classifiers.

Conclusion

Novel gene sets defined on the basis of regulatory interactions improve set-level classification of gene expression data. The experimental scripts and other material needed to reproduce the experiments are available at http://ida.felk.cvut.cz/novelgenesets.tar.gz.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-015-0786-7) contains supplementary material, which is available to authorized users.

The use of EST expression matrixes for the quality control of gene expression data

EST expression profiling provides an attractive tool for studying differential gene expression, but cDNA libraries' origins and EST data quality are not always known or reported. Libraries may originate from pooled or mixed tissues; EST clustering, EST counts, library annotations and analysis algorithms may contain errors. Traditional data analysis methods, including research into tissue-specific gene expression, assume EST counts to be correct and libraries to be correctly annotated, which is not always the case. Therefore, a method capable of assessing the quality of expression data based on that data alone would be invaluable for assessing the quality of EST data and determining their suitability for mRNA expression analysis. Here we report an approach to the selection of a small generic subset of 244 UniGene clusters suitable for identification of the tissue of origin for EST libraries and quality control of the expression data using EST expression information alone. We created a small expression matrix of UniGene IDs using two rounds of selection followed by two rounds of optimisation. Our selection procedures differ from traditional approaches to finding "tissue-specific" genes and our matrix yields consistency high positive correlation values for libraries with confirmed tissues of origin and can be applied for tissue typing and quality control of libraries as small as just a few hundred total ESTs. Furthermore, we can pick up tissue correlations between related tissues e.g. brain and peripheral nervous tissue, heart and muscle tissues and identify tissue origins for a few libraries of uncharacterised tissue identity. It was possible to confirm tissue identity for some libraries which have been derived from cancer tissues or have been normalised. Tissue matching is affected strongly by cancer progression or library normalisation and our approach may potentially be applied for elucidating the stage of normalisation in normalised libraries or for cancer staging.

Cholesterol synthesis-related enzyme oxidosqualene cyclase is required to maintain self-renewal in primary erythroid progenitors

OBJECTIVES

Molecular mechanisms controlling cell fate decision making in self-renewing cells are poorly understood. A previous transcriptomic study, carried out in primary avian erythroid progenitor cells (T2ECs), revealed that the gene encoding oxidosqualene cyclase (OSC/LSS), an enzyme involved in cholesterol biosynthesis, is significantly up-regulated in self-renewing cells. The aim of the present work is to understand whether this up-regulation is required for self-renewal maintenance and what are the mechanisms involved.

MATERIALS AND METHODS

To investigate OSC function, we studied effects of its enzymatic activity inhibition using Ro48-8071, a specific OSC inhibitor. In addition, we completed this pharmacological approach by RNAi-mediated OSC/LSS knockdown. The study of OSC inhibition was carried out on both self-renewing and differentiating cells to observe any state-dependent effect.

RESULTS

  Our data show that OSC acts both by protecting self-renewing T2EC cells from apoptosis and by blocking their differentiation program, as OSC inhibition is sufficient to trigger spontaneous commitment of self-renewing cells towards an early differentiation state. This is self-renewal specific, as OSC inhibition has no effect on erythroid progenitors that have already differentiated.

CONCLUSIONS

Taken together, our results suggest that OSC/LSS expression and activity are required to maintain cell self-renewal and may be involved in the self-renewal versus differentiation/apoptosis decision making, by keeping cells in a self-renewal state.

hSAGEing: an improved SAGE-based software for identification of human tissue-specific or common tumor markers and suppressors

BACKGROUND

SAGE (serial analysis of gene expression) is a powerful method of analyzing gene expression for the entire transcriptome. There are currently many well-developed SAGE tools. However, the cross-comparison of different tissues is seldom addressed, thus limiting the identification of common- and tissue-specific tumor markers.

METHODOLOGY/PRINCIPAL FINDINGS

To improve the SAGE mining methods, we propose a novel function for cross-tissue comparison of SAGE data by combining the mathematical set theory and logic with a unique "multi-pool method" that analyzes multiple pools of pair-wise case controls individually. When all the settings are in "inclusion", the common SAGE tag sequences are mined. When one tissue type is in "inclusion" and the other types of tissues are not in "inclusion", the selected tissue-specific SAGE tag sequences are generated. They are displayed in tags-per-million (TPM) and fold values, as well as visually displayed in four kinds of scales in a color gradient pattern. In the fold visualization display, the top scores of the SAGE tag sequences are provided, along with cluster plots. A user-defined matrix file is designed for cross-tissue comparison by selecting libraries from publically available databases or user-defined libraries.

CONCLUSIONS/SIGNIFICANCE

The hSAGEing tool provides a combination of friendly cross-tissue analysis and an interface for comparing SAGE libraries for the first time. Some up- or down-regulated genes with tissue-specific or common tumor markers and suppressors are identified computationally. The tool is useful and convenient for in silico cancer transcriptomic studies and is freely available at http://bio.kuas.edu.tw/hSAGEing.