Improving Missing Value Imputation in Microarray Data
Author Information
Author(s): Peter Johansson, Jari Häkkinen
Primary Institution: Lund University
Hypothesis
Can incorporating spot quality weights improve the accuracy of missing value imputation methods for microarray data?
Conclusion
Including a measure of spot quality improves the accuracy of the missing value imputation.
Supporting Evidence
- Weighted methods performed better than non-weighted methods.
- The weighted nearest neighbours method (WeNNI) achieved the best performance and robustness.
- Using continuous weights makes imputation less sensitive to parameter choices.
Takeaway
This study shows that using quality weights when filling in missing data in gene studies makes the results more accurate.
Methodology
The study compared several imputation methods on three datasets containing replicate measurements, focusing on weighted methods versus non-weighted methods.
Potential Biases
The study acknowledges potential biases in the experimental design affecting the imputation methods.
Limitations
The findings are based on comparisons of replicate data, which may not be available in every experimental setting.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website