SKCCC Bioinformatics Shared Resource

The last decade has seen a substantial growth in the use of high-throughput molecular technologies in cancer research across the SKCCC. Efficient utilization of the data generated by these experiments is of strategic scientific importance and requires bioinformatics support. The bioinformatics shared resource guarantees the availability of comprehensive bioinformatics expertise to Cancer Center members. This resource comprises faculty and support staff able to support data acquisition (including study design, feasibility of objectives, availability of public-access genomic information, data storage, and data annotation), statistical quality control (including artifact detection, preprocessing, and normalization of data from genomic technologies), data analysis (including visualization, modeling, inference and interpretation), and development of innovative customized bioinformatics tools, and education. This resource stabilizes and expands existing high-quality expertise in the areas of computational molecular biology, bioinformatics, and computing-intensive statistical genetics. Organizing this expertise as a Shared Resource is a cost-effective approach to ensure that qualified bioinformatics support is readily available to investigators from all programs in the center. Resource members are in an ideal position to initiate and promote interdisciplinary interactions among cancer research projects led by different investigators, and thus speed the bi-directional exchange between basic and clinical science. Members of the Shared resource will also continue to develop their own agenda of cancer bioinformatics research, to participate in educational activities in the Cancer Center and across the University, and to be active in the profession.

Experimental Design

Investigators: M Kortenhorst (Carducci lab), G Parmigiani (BISR) W Yu (Microarray core). Goal: two-channel microarray experimental design to efficiently investigate the patterns of resistance of two drugs on two genetically different cell lines at two different time points. Figure: the custom-made, “double-cube” design that was developed to incorporate consideration of scientific goals as well as specific constraints of the project.

Data Analysis

Investigators: AJ Mamelak (Sauder Lab) J Kowalski (SBIR) A Blackford (SBIR) M Zahurak (SBIR). Goal: To select genes that are differentially expressed between tissues but with greater representation from the tumor area of interest, based on samples in which the area of interest (tumor) is excised along with surrounding areas of normal skin. This is reflected in expression measurements and needs to be handled with special methods. Figure: Image plots of the similarity measure for gene expressions in BCC tumors versus normal skin. Each block represents the degree of similarity in log2-transformed intensity ratios between a pair of two samples, ranging from -0.6 to 1.0. Experiment (Exp) 1 represents an initial array and Exp 2 represents the reciprocal-labeling study. The upper, right triangular portions of the top plots represent the observed similarity between sample pairs, among all genes. The heterogeneity inherent to the tumor samples makes gene expression profiling and candidate selection challenging. In contrast, the similarity between sample pairs among a set of five candidate genes selected by the CAM method (upper triangles, bottom plots) is greater, indicating more homogeneity among the samples. The CAM method (Correlative analysis of microarrays) was developed specifically to address the needs of this challenging analysis (Mamelak et al 2005). Through this collaboration Dr. Kowalski and colleagues developed an appraoch to more generally address microarray analyses of tumors that characteristically display dysregulated growth and possess heterogeneous genetic compositions.