Comparing Methods for Identifying Differentially Expressed Genes
Author Information
Author(s): Jeffery Ian B, Higgins Desmond G, Culhane AedÃn C
Primary Institution: University College Dublin
Hypothesis
How do different feature selection methods compare in identifying differentially expressed genes from microarray data?
Conclusion
The choice of feature selection method significantly affects the classification success of gene lists derived from microarray data.
Supporting Evidence
- Only 8 to 21% of genes were in common across all 10 feature selection methods.
- Area under the ROC curve performed well with datasets that had low levels of noise and large sample size.
- Rank products performed well when datasets had low numbers of samples or high levels of noise.
- The empirical bayes t-statistic performed well across a range of sample sizes.
Takeaway
This study looked at different ways to find important genes in data from experiments. It found that the method you choose can really change the results.
Methodology
The study compared 10 feature selection methods applied to 9 publicly available binary microarray datasets.
Potential Biases
Potential biases may arise from the limited number of samples in the datasets used.
Limitations
The study's findings may not generalize to datasets with different characteristics or larger sample sizes.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website