Application of representational microarrays to genetic analysis of primary tumours

We have explored using DNA microarrays for measuring gene copy number in cancer biopsies. The high nucleotide complexity of the human genome makes hybridization analysis of the entire genome problematic. Therefore, we have used the approach of making complexity-reducing representations¹. Accordingly, the printed probes themselves are derived from simplified representations of the human genome, and are hybridized to representations of tumour and normal DNA. Data from pilot experiments indicate that this method can be used to detect amplifications and deletions in cancer specimens. The theoretical resolving power of these microarrays is on the order of 100 kb with a 30,000-member array, but higher resolving power is not excluded. In contrast to expression array analysis, only minute amounts of starting material (5,000 nuclei) are sufficient for sample preparation. Also, the starting material for analysis is DNA not RNA, which is easier to obtain from surgical specimens. The two methods should work best in concert; that is, combining copy-number analysis with expression analysis for the same tumours, enabling the user to more readily identify the primary defects in cancer cells. Presently we have produced a 2,000-member array that contains fragments of known sequence and mapping location as well as fragments cloned randomly from a representation. The copy number of many of these probes is known in respect to several tumours we have characterized by more conventional methods. Representations from these tumour pairs have been hybridized to these arrays and analysed. By querying the hybridization results we have identified the fragments known to be effected in these tumours. We are now confirming the results of the hybridizations by more accepted techniques. Several obstacles remain to creating a high-density genomic chip, including the assembly of a collection of probes, mapping them and developing the software tools to seamlessly analyse the data. Genomic chips have uses beyond cancer, including detection of polymorphisms in individuals and de novo germline rearrangements.