Improving Gene Discovery with Semi-Supervised Methods
Author Information
Author(s): Oba Shigeyuki, Shin Ishii
Primary Institution: Nara Institute of Science and Technology
Hypothesis
Can the optimal discovery procedure (ODP) framework improve the detection of active genes in unsupervised and semi-supervised settings?
Conclusion
The ODP framework is effective for detecting active genes and is enhanced by sharing estimations of hidden variables across multiple tests.
Supporting Evidence
- The ODP framework improves detectability by sharing estimated parameters.
- Unsupervised and semi-supervised scores can enhance gene ranking stability.
- ODP-SS outperformed other methods in specificity for detecting active genes.
Takeaway
This study shows that using both labeled and unlabeled samples can help find important genes better than using just labeled samples.
Methodology
The study used a latent variable model and the optimal discovery procedure (ODP) to analyze gene expression data.
Potential Biases
Labeling errors may affect the accuracy of gene detection.
Limitations
The study did not consider false discovery rate (FDR) estimation in unsupervised cases due to missing class labels.
Participant Demographics
The study involved prostate and colon cancer samples.
Statistical Information
P-Value
p<0.001
Statistical Significance
p<0.001
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website