CSIRO’s new statistical methods enable analysis of very complicated microarray data.

Smart statistics for bioinformatics

CSIRO is developing new statistical tools for analysing microarray data to enable faster drug discovery and development of simpler clinical diagnostic tests.

• 28 January 2011 | Updated 14 October 2011

• The issue
• What we did
• Outcomes
• Simpler clinical diagnostics
• Toxicogenomics
• Pharmacogenomics

The issue

Microarray experiments generate the sort of data where the number of measurements of each sample is much greater than the number of samples.

These massively multivariate datasets are unable to be analysed by traditional statistical methods.

What we did

CSIRO developed GeneRave, a new statistical technique specifically for microarray data.

GeneRave is able to cope with the multivariate nature of microarray data and extract meaningful information from it.

GeneRave is also able to handle other types of multivariate data, including protein expression data and single nucleotide polymorphism (SNP) data.

Outcomes

GeneRave can be used for developing:

• simpler clinical diagnostics
• efficient screening methods for potential drug candidates - ‘toxicogenomics’
• genetic tests for drug efficacy - ‘pharmacogenomics’.

Simpler clinical diagnostics

GeneRave enables us to identify diagnostic markers that are based on a very small number of genes.

The fewer genes that are required to diagnose a disease, the simpler and cheaper the diagnostic tests can be.

Diagnostic tests requiring only a small number of genes can use existing platforms rather than expensive microarray-based platforms.

For example, GeneRave is able to accurately classify patients with diffuse large B-cell lymphoma into two distinct subgroups by analysing the expression of just two genes.

Other technologies require many more genes for classification.

GeneRave's accurate classification could lead to a simple diagnostic test and a more effective treatment regimen.

Toxicogenomics

Shortcuts in drug discovery have the potential to save drug companies millions of dollars and cut months or years off the time taken to get new drugs onto the market.

Using GeneRave to analyse microarray data from bacteria treated with drugs that have known modes of action, we discovered that the gene expression profile of the bacteria was indicative of the mode of action of the drug.

This means that our statistical techniques may enable determination of the mode of action of a candidate drug from simple microarray experiments.

We hope this will lead to more efficient methods for screening candidate drugs to treat a range of diseases.

Pharmacogenomics

The ability to identify subpopulations that would benefit from a particular drug (or suffer a significant side effect) using their genetic makeup would reduce the number of drugs failed by regulators due to their effects on small segments of the population.

It would also enable advances in personalised medicine.

Our interest in this area centres on the relationship between a patient’s SNP profile, or a tumour’s expression profile, and the patient’s response to a particular treatment.

Learn more about CSIRO's work in Bioinformatics.