Shili Lin, The Ohio State University

Title: Class Discovery and Classification of Tumor Samples Using Mixture Modeling of Gene Expression Data - a Unified Approach

Abstract:
Motivation: The DNA microarray technology has been increasingly used in cancer research. In the literature, discovery of putative classes and classification to known classes based on gene expression data have been largely treated as separate problems. This article offers a unified approach to class discovery and classification, which we believe is more appropriate, and has greater applicability, in practical situations.

Results: We model the gene expression profile of a tumor sample as from a finite mixture distribution, with each component characterizing the gene expression levels in a class. The proposed method was applied to a leukemia dataset, and good results are obtained. With appropriate choices of genes and preprocessing method, the number of leukemia types and subtypes is correctly inferred, and all the tumor samples are correctly classified into their respective type/subtype. Further evaluation of the method was carried out on other variants of the leukemia data and a colon dataset.

This is joint work with Roxana Alexandridis and Mark Irwin.

Tomas Radivoyevitch, Case Western Reserve University

Title: Models of Leukemia

Abstract: This talk will consist of two parts. In the first part I will describe my previous research in biologically-based models of radiation-induced chronic myeloid leukemia. In the second part I will describe my preliminary results toward models for improved therapy of BCR-ABL childhood acute lymphoblastic leukemia.

Mike West, Duke University

Title: Statistical Prediction Tree Modelling in Clinico-Genomics

Abstract: This talk will review some of the models and methods underlying our work in clinico-genomics. A key focus is on the development of integrated models that include molecular profiles(DNA microarray based gene expression data) together with traditional clinical and pathological factors in prognostic models. Our work in breast cancer provides example contexts, some relevant experiences and stimuli for open questions and challenges. The talk will focus on modelling and predicting cancer outcomes, and cover topics related to our development of Bayesian predictive classification tree analyses - aspects of model development and fitting, and current challenges to methodology, as well as applications. Broader developments will be discussed, including modelling developments related to factor analysis For large-scale gene expression patterns that aim in part, at improved characterisation of molecular profiles for inclusion in predictive models; other issues and current foci include development of stochastic computation for prediction tree modelling. We will also touch on issues arising as such clinico-genomic studies move towards clinical evaluation and eventual implementation to begin to realise the promise of large-scale molecular data in personalized health care.

Xuejun Peng, Cleveland Clinic Foundation

Title: Statistical analysis of comparative genomic hybridization microarray data

Abstract: Comparative genomic hybridization array (CGHa) is a technology that enables investigators to identify presence of chromosome changes (i.e. gains and losses of chromosomes and chromosome regions) directly from DNA samples. Its application has seen rapid growth in cancer studies. While it shares many characteristics of 2-color cDNA gene expression arrays, this technology brings many unique statistical analysis issues from data pre-processing to bioinformatics interpretation. In this short talk, we will discuss some of these issues using real data from a well-replicated CGHa experiment.