Title
Gene expression profiling for drug discovery and Selection of Therapy
Speaker
John Weinstein,, Group Chief, Laboratory of Molecular Pharmacology, National Cancer Institute
Abstract
It has proved easier to use mRNA expression profiles for classification of tumors and for prognosis than it has been to integrate such profiles into the drug discovery process. To do so, we and our collaborators have used both cDNA microarrays and oligonucleotide chips to characterize patterns of gene expression in 60 human cancer cell lines used by the National Cancer Institute's drug discovery program. Since those cancer cells have also been characterized by their patterns of sensitivity to >70,000 chemical agents, we are afforded a unique opportunity to integrate gene expression with pharmacology and toxicology. To facilitate mining of these large databases, we have developed a number of algorithms for data analysis, pattern visualization, and literature searching. The first was the color-coded Clustered Image Map. This visualization tool compacts high-dimensional data into a two-dimensional space and brings out patterns of association. A second tool, MedMiner, speeds up 5- to 10-fold the process of searching the biomedical literature for meaningful gene-gene and gene-drug relationships. One result with possible clinical implications has been the finding of high correlations between expression of the gene asparagine synthetase and potency of the enzyme-drug L-asparaginase. As we enter what might be termed the "omic" era (i.e., genomic, proteomic, clinomic, functional genomic, etc.), integrated pharmacogenomic databases such as these will have a great impact on therapy, both by accelerating the drug discovery process and by providing a molecular basis for individualization of therapy.
