Prediction potential of candidate biomarker sets identified and validated on gene expression data from multiple datasets
2007

Predicting Cancer Biomarkers from Gene Expression Data

Sample size: 1000 publication 10 minutes Evidence: moderate

Author Information

Author(s): Gormley Michael, Dampier William, Ertel Adam, Karacali Bilge, Tozeren Aydin

Primary Institution: School of Biomedical Engineering, Drexel University, Philadelphia, PA, USA

Hypothesis

Can gene expression profiles accurately predict molecular phenotypes and relapse in cancer?

Conclusion

Many gene sets can accurately predict molecular phenotypes, but predicting relapse from microarray data is challenging.

Supporting Evidence

  • The study found that supervised feature selection led to more accurate classifications than random selection.
  • Expression profiles correlated with molecular phenotypes were identified across multiple datasets.
  • Prediction error for relapse-free status was consistently high across different cancer types.

Takeaway

Scientists looked at gene data to find patterns that can help predict cancer types and outcomes, but it's hard to guess if someone will relapse just from this data.

Methodology

The study used supervised univariate feature selection algorithms on multiple microarray datasets to create expression profiles.

Potential Biases

Potential overfitting due to the use of leave-one-out cross-validation and variability in sample populations.

Limitations

The study faced challenges in predicting relapse due to high prediction errors and variability in gene expression across datasets.

Participant Demographics

Included patients with breast cancer, lymphoma, and renal cancer from multiple datasets.

Statistical Information

P-Value

p<0.05

Confidence Interval

95% CI

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-8-415

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication