ABS04 - 2004 Applied Bayesian Statistics School

STATISTICS & GENE EXPRESSION GENOMICS:
METHODS AND COMPUTATIONS

Centro Congressi Panorama, Trento, Italy

15 - 19 June, 2004

Centro Congressi Panorama



PROGRAMME

**********************************
June 15th - Lecture 1: Overview
**********************************

Genome-scale data and biological investigations

Examples of molecular phenotyping in areas such as cancer

Statistical principles and basics

Molecular biology principles and basics

DNA microarray basics and technologies 

Statistics resources for DNA microarray data

Basic data issues: management, imaging, normalisation, other

PRACTICAL SESSION

Basics of managing, manipulating and exploring gene expression
data sets in R and Matlab

Web resources for gene and genomic information,
in particular Bioconductor resources for R

Examples: various data sets, with major focus on Affymetrix microarrays

*******************************************************************************
June 16th - Lecture 2: Aspects of oligonucleotide array
                       data processing and summary
*******************************************************************************

Expression intensity models

Aspects of exploratory data analysis with examples selected from studies in
cancer phenotyping, cell cycle studies, designed experiments in mouse models,
cardiovascular disease genomics and other

Exploratory uses of regression - collinearities among genes 

Data reduction - gene subset selection and screening
principal components/singular factors 

Interactive illustrations and examples (Matlab)

Images, clustering

PRACTICAL SESSION

Exploratory data analysis using example data sets (Matlab, R with Bioconductor)

Normalisation, screening and computing expression intensity estimates

Linking analysis results to gene ontologies and biological literatures

Clustering examples and tools (R, xcluster, cluster/treeview)

*******************************************************************************
June 17th - Lecture 3: Regression modelling with many predictors
*******************************************************************************

Gene expression as covariate information

Bayesian analysis of linear and binary principal component/factor regression

Some theory of Bayesian actor models and related topics

Relevance of shrinkage priors - generalised g-priors

Phenotyping examples in breast cancer, cell line and animal studies

Trans-species molecular phenotyping: cancer signatures

Case control vs prospective studies

Metagenes and data summaries for predictive modelling

*******************************************************************************
June 18th - Lecture 4: Bayesian predictive regression tree models
*******************************************************************************

Binary response prediction using Bayesian tree models
for retrospective sampling
Examples in cardiovascular disease prediction, breast cancer risk prediction

More general statistical tree models: survival analysis
using classification trees 
Use of metagene summaries in predictive models
Examples of recurrence prediction
Clinico-genomic models 

Exploration of binary tree software (Matlab)

PRACTICAL SESSION

Fitting regression models in Matlab and R

Exploring gene subset predictor selection 

Bayesian factor/binary regression explorations

More on Bioconductor tools for data processing
and expression intensity estimation

*******************************************************************************
June 19th - Lecture 5: 	Graphical models for gene expression data
                        representation, visualisation and exploration
*******************************************************************************

Theory of graphical models and compositional networks

Ideas of model fitting using directed graphs

Bayesian analysis and theory - prior specification, posterior computation

Large-scale regression model search - "Shotgun Stochastic Search"
and its uses in exploring and evaluating large-scale graphical models

Examples from breast cancer, cell growth and signalling gene pathways,
and other areas

Illustrations of software (GraphExplore) for exploring and manipulating graphs