Multivariate hierarchical Bayesian model for differential gene expression analysis in microarray experiments

Biological assessment of robust noise models in microarray data analysis

Motivation: Although several recently proposed analysis packages for microarray data can cope with heavy-tailed noise, many applications rely on Gaussian assumptions. Gaussian noise models foster computational efficiency. This comes, however, at the expense of increased sensitivity to outlying observations. Assessing potential insufficiencies of Gaussian noise in microarray data analysis is thus important and of general interest.

Results: We propose to this end assessing different noise models on a large number of microarray experiments. The goodness of fit of noise models is quantified by a hierarchical Bayesian analysis of variance model, which predicts normalized expression values as a mixture of a Gaussian density and t-distributions with adjustable degrees of freedom. Inference of differentially expressed genes is taken into consideration at a second mixing level. For attaining far reaching validity, our investigations cover a wide range of analysis platforms and experimental settings. As the most striking result, we find irrespective of the chosen preprocessing and normalization method in all experiments that a heavy-tailed noise model is a better fit than a simple Gaussian. Further investigations revealed that an appropriate choice of noise model has a considerable influence on biological interpretations drawn at the level of inferred genes and gene ontology terms. We conclude from our investigation that neglecting the over dispersed noise in microarray data can mislead scientific discovery and suggest that the convenience of Gaussian-based modelling should be replaced by non-parametric approaches or other methods that account for heavy-tailed noise.

Contact:peter.sykacek@boku.ac.at

Availability:http://bioinf.boku.ac.at/alexp/robmca.html.

Microarray bioinformatics

Bioinformatics has become an increasingly important tool for molecular biologists, especially for the analysis of microarray data. Microarrays can produce vast amounts of information requiring a series of consecutive analyses to render the data interpretable. The direct output of microarrays cannot be directly interpreted to show differences in settings, conditions of samples, or time points. To make microarray experiments interpretable, it is necessary that a series of algorithms and approaches be applied. After normalization of generated data, which is necessary to make a comparison feasible, significance analysis, clustering of samples and biological compounds of interest and visualization are generally performed. This chapter will focus on providing a basic understanding of the generally approaches and algorithms currently employed in microarray bioinformatics.

Bioinformatics research in the Asia Pacific: a 2007 update

We provide a 2007 update on the bioinformatics research in the Asia-Pacific from the Asia Pacific Bioinformatics Network (APBioNet), Asia's oldest bioinformatics organisation set up in 1998. From 2002, APBioNet has organized the first International Conference on Bioinformatics (InCoB) bringing together scientists working in the field of bioinformatics in the region. This year, the InCoB2007 Conference was organized as the 6th annual conference of the Asia-Pacific Bioinformatics Network, on Aug. 27-30, 2007 at Hong Kong, following a series of successful events in Bangkok (Thailand), Penang (Malaysia), Auckland (New Zealand), Busan (South Korea) and New Delhi (India). Besides a scientific meeting at Hong Kong, satellite events organized are a pre-conference training workshop at Hanoi, Vietnam and a post-conference workshop at Nansha, China. This Introduction provides a brief overview of the peer-reviewed manuscripts accepted for publication in this Supplement. We have organized the papers into thematic areas, highlighting the growing contribution of research excellence from this region, to global bioinformatics endeavours.