Multidimensional gene expression profiling
Mikael Kubista, Department of Chemistry and Biosciences

During the last decades pharmaceutical industry has been very successful developing drugs to cure diseases and relieve symptoms. For most diseases of lower complexity, the health care system in the Western world can today offer acceptable treatments, but for complex diseases cures remain remote. With complex diseases it is not sufficient to hit one biological target, but one must moderate the interplay between biological components, which in turn requires interacting with many targets in a well tuned manner. This is of course much more complicated. In most cases we don't even know which these factors are, and how the potential drug targets in a complex disease interact, and appropriate tools to find this out have not been available.

A new technology, called real-time PCR, may provide the necessary platform to address these challenges. By real-time PCR the amount of a particular nucleic acid can be determined with the accuracy of +/- 5 % and sensitivity sufficient to detect a single molecule in a complex biological sample. The technology was developed by a few independent groups during the last decade, one of which was the applicant's laboratory. Kubista developed the light-up probes, around which patent portfolio LightUp Technologies AB was founded as the worlds' first company specialized in real-time PCR based diagnostics. Today LightUp products are used in routine at hospitals all over the world. Later TATAA Biocenter was founded as world's first service provider of real-time PCR services, and was soon established as the prime workshop organizer and research contractor in real-time PCR.

Next goal is to develop a technology platform to identify potential targets to treat complex diseases, and tools to investigate how they interact on molecular level and respond to environmental factors, drug candidates, and treatment. The following strategy will be used: By real-time PCR we can measure the expression of genes with unprecedented accuracy in complex biological samples. We can even measure expression in individual cells collected by laser capture microdissection. This makes it possible to monitor how rare cells in complex tissues are affected by environment, drugs or simply how they develop in time. These highly accurate real-time PCR expression profiles reflect the detailed conditions of the studied tissues. By monitoring how the expression profiles change with time, we can monitor disease progress, healing, and study effects of genetic and environmental factors, including the effects of chemicals and drug candidates. The collected data will be multidimensional and multivariate, with expression data of many genes, in different tissues in animal models measured as function of time, and optionally also as function of environmental conditions, animals' age, drug load etc. From these data, information about the interaction of different factors will be extracted, which will be of uttermost value to understand the biology of the studied disease and the effects of treatments and drugs.

To extract information and to identify significant correlations in the multidimensional data sets, advanced multidimensional analysis and clustering methods will be used. These will include multidimensional principal component analysis, generalized Procrustes rotation, PARAFAC, neural networks, support vector machines, genetic algorithms, and artificial intelligence. The latter methods are particularly interesting since they can find also non-linear correlations, such as: expression of gene a increases two weeks after treatment with at least b mg of drug, resulting in improved healing in individuals of a specific genotype. The project will be performed in collaboration with MultiD Analysis AB, SAIDA Medical AB and the Bristol Center of Chemometrics.