Cancer classification with DNA microarrays - is less more?

The dissection of cancer and the underlying molecular processes that are de fective in cancer cells has become an important tool in the fight against t his disease. DNA microarrays can provide detailed information of the expres sion pattern of thousands of genes in tumours. But how much of this data is useful and is some superfluous? Can array data be used to identify a handf ul of critical genes that will lead to a more-detailed taxonomy of tumours and can this or similar array data be used to predict clinical outcome? Pri mary tumours will give us the statistical power to draw these conclusions, but can cancer cell lines be used as models to point us in the right direct ion?