Gene Expression profiling

DNA arrays have become the preferred method for large-scale measurement of gene expression. Expression profiling using DNA arrays is a step in the direction of functional characterisation and can almost be performed on the required scale. There are several different implementations of the DNA array principle for expression measurement. Miniaturised devices, glass microarrays and oligonucleotide chips, are the most promising in terms of throughput and should eventually allow simultaneous measurement of expression on the whole set of 'Unigene-defined' human genes. Nylon macroarrays (high-density filters) continue to be used because of their accessibility and flexibility. In fact, they enjoy high demand as academic laboratories discover that glass microarrays and oligonucleotide chips are somewhat inaccessible. Finally, Nylon microarrays, with both colourimetric and radioactive probes, offer a promising alternative in a number of situations. In particular, where detection of genes expressed at the 1/100,000 level is required, radioactive probing of Nylon microarrays can provide similar results with very small amounts of sample: this is very significant for a number of research and clinical applications.

While microarrays and chips have received wide publicity, and the pharmaceutical industry has made a significant investment in this area, academic scientists may be left with the impression that the technology is incredibly powerful but prohibitively expensive. Since a major bottleneck for the academic community is the set-up cost, the programme will provide a network through which the technology can be made more accessible. Many groups will wish to manufacture their own arrays, representing sets of genes expressed in their preferred organism, tissue or pathology. Access to the cDNA clones required for preparation of the PCR products is in principle solved by the availability of the extensive set of sequenced IMAGE clones. However, sequence verification, made necessary by a 10-20% error rate in this collection, as well as the logistics of producing thousands of PCR products, are not simple tasks and are only two of the barriers preventing smaller research groups from using expression profiling technology. If a system can be organised such that laboratories with minimal equipment can readily exploit expression profiling technology through cooperation, then suborganisation could be also implemented by research field, e.g. yeast, human, mouse, genetic disease, cancer, immunology, etc. Cooperation can be developed at various levels depending on the expertise and instrumentation available in different centres. It will require a powerful, user-friendly database system, standardisation of correction and normalisation procedures so that data points from different projects are indeed comparable, and a common willingness to share data.

Data handling itself is a major issue: users can be quickly swamped by tens of thousands of measurements and may not be able to handle them, much less extract the information they contain. A well-organised and coherent software suite is necessary to move from raw data to corrected and normalised expression values, and imaginative statistical and representation tools must be implemented to allow the biologist to look at data in as many ways as possible, while integrating information available on the internet for the same set of genes. Data archiving is a related but distinct issue. Indeed, the same arrays are normally used by several groups, and by several individuals within a single laboratory. It therefore makes sense to ensure that all the data, in a suitably standardised form, is available to each participant in some kind of laboratory notebook system. An amenable and effective database system is clearly essential, together with appropriate standardisation to ensure that data from different laboratories can be compared.