Clustering of log-fold change with zeros

Question

[boczniak767]

Hi all,

I have made a table of log2 fold change ratios from several microarray studies. For genes without statistically significant change I typed 0 (there are many such cases), other values are just log2 fold changes.

My question is, if it is possible to make clustering of such data? I could, of course, use raw values but I wanted to include only genes with significantly changed expression.

I have tried heatmap in R, it works but if it is sensible?

[ATpoint]

Where is the problem? Why don't you simply use only the significant ones? Please elaborate.

[boczniak767]

The problem is that the majority of data are zeros. There are no (or very few) rows (genes) with significant data for all columns (projects). Full set has ca. 10 columns, I could use smaller set of five, but the problem remains.