# BakaWiki

### Site Tools

topics:bib:laur2011c

# Binarization of expression data

 Category: Data Analysis Raivo Kolde 2011 Binarization of expression data can be beneficial if we are trying to aggregate data over various experiments from different labs on various microarray platforms. Using binarization can increase robustness by removing noise and retaining only the most essential signal. The goal is to find methods for the task and test these on actual data. The criterions for testing are classification accuracy using samples from different experiments and correlation over different microarray platforms. {{}}

* Get an overview of the normalization of Affymetrix microarrays, especially the expression detections algorithms like MAS5.0 and fRMA.
• Select MEM datasets from different platforms that represent similar biological conditions
• Implement/run the binarization algorithms on these datasets
• Evaluate the goodness of methods, by comparing
• the classification accuracy, when learning happens in one set of datasets and evaluation on other
• how similar are gene profiles on similar samples when different microarray platforms are used

# Literature

• McCall et al. Frozen robust multiarray analysis (fRMA). Biostatistics (2010) vol. 11 (2) pp. 242-53