The first part of the talk concerns detection of alternative splicing using Affymetrix exon tiling arrays. Alternative splicing is known to be a critical factor in cancer formation and progression. In real experiments, high heterogeneity is often observed among cancer patients. Specifically, alternative splicing variants may show different degrees among or only occur to subgroups of cancer patients. We propose a penalized mixture statistical model integrated with dimension reduction of the interaction space based on ANOVA-type model and a sequential testing procedure to detect genes with such cancer subgroup structure.Discovered important biomarkers can potentially be used to guide personalized medicine. In the second part of the talk, I will introduce a new family of covariate-adjusted response-adaptive (CARA) randomization procedures based on two important factors of efficiency and medical ethics. It is capable of incorporating covariate information such as the genetic profile in the response-adaptive randomization of clinical trials.
Return to Biostatistics Seminars