Improving Missing Value Imputation in Microarray Data
Author Information
Author(s): Cecilia Ritz, Patrik Edén
Primary Institution: Lund University, Sweden
Hypothesis
Can accounting for one-channel depletion improve the imputation of missing values in 2-dye microarray data?
Conclusion
By including more information in the imputation step, we more accurately estimate missing expression values.
Supporting Evidence
- KNN-based imputation showed a systematic bias for one-channel depleted spots.
- Mean square deviation between imputed values and duplicates was reduced up to 51%.
- The method improved imputation for one-channel depleted spots across five datasets.
Takeaway
This study shows that when some data is missing in one part of a test, we can use information from the other part to make better guesses about what the missing data should be.
Methodology
The study used KNN-based imputation and cross-validation to evaluate the correction of bias in imputed values.
Potential Biases
The study acknowledges potential biases in conventional imputation methods.
Limitations
The method does not apply to experiments where the two dyes are used for pairwise comparison of biological traits.
Participant Demographics
The datasets included samples from breast cancer, lymphoma, melanoma, and mycoepidermoid carcinoma.
Statistical Information
P-Value
p<0.05
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website