A Bayesian calibration model for combining different pre-processing methods in Affymetrix chips
2008

A Bayesian Model for Combining Pre-processing Methods in Gene Expression Studies

Sample size: 10621 publication Evidence: high

Author Information

Author(s): Marta Blangiardo, Sylvia Richardson

Primary Institution: Centre for Biostatistics, Imperial College, London, UK

Hypothesis

Can a Bayesian calibration model improve the assessment of differential expression in gene expression studies by combining multiple pre-processing methods?

Conclusion

The Bayesian calibration model outperforms individual pre-processing methods in assessing differential expression.

Supporting Evidence

  • The Bayesian model showed improved sensitivity and specificity in detecting differentially expressed genes compared to individual methods.
  • Using the combined model, 292 probesets were identified as differentially expressed, significantly more than any single method.
  • The model effectively synthesizes information from multiple pre-processing methods, enhancing the biological interpretability of results.

Takeaway

This study shows that using a special math model can help scientists get better results when looking at gene data by combining different ways of preparing that data.

Methodology

The study used a Bayesian calibration model implemented in WinBUGS to analyze gene expression data from Affymetrix chips, comparing the performance of multiple pre-processing methods.

Potential Biases

The study acknowledges potential biases from the choice of pre-processing methods.

Limitations

The model's applicability is limited to differential expression assessment and relies on having generally applicable pre-processing methods.

Digital Object Identifier (DOI)

10.1186/1471-2105-9-512

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication