we feel that CRC provide some unique advantages and additional functions over other model-based clustering 
methods. The improvements are summarized as follows: 

1.	Cluster genes showing non-synexpression correlation patterns (time-shifted and/or inverted) together 
in the same group. This feature potentially allows more functionally related genes to be recognized.
2.	Provide accurate estimation of number of clusters. Sometimes such knowledge can be quite useful. For 
example, in the yeast galactose data example, CRC is able to accurately estimate the number of GO functional 
categories. 
3.	Automatically andle missing data during clustering. No preprocessing elimination or 
imputation step is needed.
4.	Provide multiple strength measurements for each cluster produced, including tightness measure and 
stability measure. Such information can help investigators to generate high quality hypotheses and identify 
significant clusters to perform follow up studies.   

We believe these advantages make CRC a highly competitive choice among all the clustering tools for analyzing 
complex microarray gene expression data.