Program | Registration | Organizers | Sponsors | ICSB2010 | Poster  


The successes, challenges and prospects for next generation GWAS analyses for complex diseases

Edinburgh, UK, 15. Oct. 2010

*UPDATE* Note that the workshop is in the Informatics Forum ( 10 Crichton Street, Edinburgh EH8 9AB ), not the main ICSB venue.


Purpose of the workshop

Genome-wide association studies (GWAS) have become the method of choice for studying disease etiology with an increasing number of GWAS studies reporting progress towards uncovering the genetic markers for complex diseases such as schizophrenia1. This workshop is aimed at introducing an interdisciplinary audience to the concept and practice of GWAS. It will cover the fundamental assumptions, showcase recent successes and discuss limitations of current GWAS approaches in the field of complex diseases. It will provide a stage for shaping the next generation of GWAS by drawing on the audience's interdisciplinary expertise in statistics and machine learning to overcome present challenges and identify the most promising avenues of future research.


GWAS has been enabled by three major advances in human genetics: the publication of the human genome, the characterisation of common human genetic variation by the International HapMap Project, and the development of SNP arrays that enable reliable and affordable screening of up to 1 million single nucleotide polymorphisms (SNPs) simultaneously. Advances in statistical methods to identify disease-associated genetic variants without bias have also been important. Despite the success for some diseases where SNPs identified by GWAS explain a substantial fraction of the genetic risk (e.g. age-related macular degeneration), for other complex diseases such as schizophrenia, the SNPs reported using current GWAS methods explain only a small proportion of the observed heritability. Recent studies hint that this "missing/hiding" heritability may be explained by hundreds or thousands of common small effect variants, in addition to a large number of rare large effect variants. Only a small fraction of this variation is likely to be tagged by the current generation of SNP arrays; future GWAS will involve a combination of next generation SNP arrays (containing 5 million or more SNPs) and whole genome sequencing. Identifying novel risk factors from such data will entail formidable challenges, not only because of the sheer quantity of variation, but because of complexity arising from allelic and locus heterogeneity, dosage and timing of gene expression, epistasis and epigenetic effects.

Machine learning algorithms, particularly those recently developed, hold great promise for the analysis of multidimensional GWAS data, but their potential has yet to be fully explored.

This workshop will consider how existing statistical and machine learning methods can be integrated in novel ways that advance the study of the genetic basis of complex disease.

Audience and Outcome

By bringing together researchers with an interest in augmenting GWAS with novel statistical approaches and machine learning technologies to study the etiology of complex diseases such as schizophrenia, the workshop aims to bridge the gap between statistics, bioinformatics and genetics. We envisage that researchers in machine learning and statistics who have an interest in complex disease genetics, as well as statistical geneticists looking for improved tools for GWAS will gain from participating in this workshop.



The program with more information about the speakers can be downloaded




8:30 am

Registration and morning tea

9:00 am


Bryan J. Mowry



Session 1: Introduction and showcase of GWAS potential


9:10 am

The theory and practice of GWAS

David Balding

9:30 am

Mega-GWAS of complex diseases

Bryan J. Mowry

9:50 am

Use of GWAS for prediction of genetic risk to disease

Naomi Wray

10:10 am

Discussion: Potential and challenges of GWAS

  • Power/sample size implications
  • How to link genomic variants with disease genes


Frank Dudbridge, Peter Holmans

10:40 am





Session 2: Next-gen sequencing data


11:00 am

The potential of next-generation sequencing to identify rare SNPs and structural variants

Kees Albers

11:20 am

The utility of family studies to identify causal variants

Douglas Blackwood

11:40 am

Discussion: Leveraging next-generation sequencing in complex disease genetics (RNA-seq, CHiP-seq, Ribo-seq)

  • Strength and weaknesses of arrays and sequencing platformsStrength and weaknesses of targeted re-sequencing and the appropriate reference sequence
  • Anticipated improvements of whole genome sequencing
  • How best to seek gene-environment interactions
  • How to study epigenetic influences in inaccessible tissue (brain)


Benjamin Pickard

12:30 pm





Session 3: Current best practice


1:30 pm

Integrating biological expert knowledge of gene function and disease pathways in GWAS

Hans van Houwelingen

1:50 pm

Integration of genetic, epigenetic and expression data

Mike Nalls

2:10 pm

Practice and outcome of bioinformatics analysis of GWAS data: success and limitations

Patrick Sleiman

3:30 pm

Discussion: Bioinformatics and the pursuit of the silver bullet

  • The importance of epistasis (gene-gene interaction)
    Filtering, wrapping and selecting which method to choose
  • De novo genome assembly benefits and drawbacks
    Non-computational improvements in next-gen sequencing data (paired-end and length increase)
  • Computational infrastructure: are we prepared for the flood?


Silke Szymczak

4:00 pm





Session 4: The road ahead


4:20 pm

Challenges in combining results of Genome Wide Association Studies

Eleazar Eskin

4:40 pm

Few samples with many features – lessons already learned from micro-array analyses

Geoff McLachlan

5:00 pm

Machine Learning for epistasis detection and analysing structured phenotypes

Karsten Borgwardt

5:20 pm

Discussion: Future directions for GWAS analysis with customized machine learning approaches and data integration

  • Signal to noise ratio in public databases
    Selection bias with filter- and wrapper-based feature selectionRobustness and over-fitting control
  • Risks of using "black-box" approaches
  • How to verify the accuracy of our findings


Frank Dudbridge

5:50 pm

Forums Discussion and Wrap-up

Bryan J. Mowry


Details of invited speakers

Bryan J. Mowry (University of Queensland, Brisbane, AU) confirmed
David Balding (University College London, UK) confirmed
Naomi Wray (Queensland Statistical Genetics Laboratories, Brisbane, AU) confirmed
Frank Dudbridge (MRC Biostatistics Unit, Cambridge, UK) confirmed
Peter Holmans (University of Cardiff, UK) confirmed
Kees Albers (Wellcome Trust Sanger Institute, UK) confirmed
Douglas Blackwood (University of Edinburgh, UK) confirmed
Benjamin Pickard (University of Strathclyde, UK) confirmed
Hans van Houwelingen (Leiden University, Netherlands) confirmed
Mike Nalls (Porter Neuroscience Research Center, USA) confirmed
Patrick Sleiman(The Children's Hospital of Philadelphia, USA) confirmed
Silke Szymczak (Universität zu Lübeck, Germany) confirmed
Eleazar Eskin (University of California, USA) confirmed
Geoff McLachlan (University of Queensland, Brisbane, Australia) confirmed
Karsten Borgwardt (Max Planck Institute for Developmental Biology and Max Planck Institute for Biological Cybernetics, Germany) confirmed

Additional information

Please note that this workshop is not overlapping but following the World Congress on Psychiatric Genetics (3-7. Oct 2010)


The online registration has now closed. The registration on the day of the workshop will be at the workshop venue, Informatics Forum (10 Crichton Street Edinburgh EH8 9AB), from 8:30 am.
View ICSC GWAS Workshop in a larger map


Prof. Bryan Mowry
The Queensland Brain Institute
The University of Queensland, Brisbane, Australia

Dr. Denis Bauer
The Queensland Brain Institute
The University of Queensland, Brisbane, Australia

Dr. Jake Gratten
The Queensland Brain Institute
The University of Queensland, Brisbane, Australia


Eppendorf Illumina    



The advertising poster is available here.

Last modified: Thu Oct 14 11:25:02 BST 2010