GEI and NCI Meeting

The Challenge of Mapping GWAS Signals: What Are the Statistical Limits of Follow-up Sequencing, Genotyping, and Haplotyping?

Hyatt Regency Hotel

Bethesda, Maryland
March 24-25, 2009

 

Goal: The goal of the meeting was to discuss statistical issues of discovery and experimental design, informed by specific examples, to develop more efficient and effective follow-up approaches for sequencing and genotyping for the genetic dissection of GWAS signals.


Subtext: How do we move from having this statement in every paper “This marker may be in linkage disequilibrium with another, uncharacterized marker that is the causal variant for the increased risk…” to “This set of variants is highly likely to include the variants in these regions that cause the increased risk…”?  (Laboratory studies will be needed to determine which variants are causal.)

Tuesday, March 24, 2009
     
6:00–6:15

Introduction, goals, and “Cone(s) of Truth”

Stephen Chanock
     
6:15–7:30

Session 1: GWAS, sequencing, and follow-up:  Where are we?
Examples of what has been found by sequencing and genotyping follow-up to GWAS. 

 

Moderators:                                                                                    

 David Hunter and Stephen Chanock
  Breast and prostate cancer studies Stephen Chanock
 

Wellcome Trust Case-Control Consortium 

 Mark McCarthy
  Epilepsy and ion channel studies Richard Gibbs
  Crohn’s disease and CNVs Steve McCarroll
  Type 2 diabetes studies    David Altshuler
     
7:30–7:45
 Break  
     
7:45–9:00

Session 2: Which GWAS signals should be followed up?
How well do single-marker, haplotype, and interaction analyses work for detecting GWAS signals? What range of significance in a GWAS merits follow-up? What factors other than level of significance should be considered (study design, power, allele frequency, effect size, pleiotropy, current functional annotation)? How well does correcting for artifacts work (genotyping biases, population stratification)? How should data from multiple populations be used? How can meta-analyses of multiple studies be used to choose signals?
 

 Moderators:

 Mark McCarthy and Elizabeth Gillanders
 

 Which analyses detect GWAS signals  
(including meta-analyses vs combined analyses)

 Gonçalo Abecasis
  Choosing regions for follow-up Michael Boehnke
   Data from multiple populations Carlos Bustamante
     
Wednesday, March 25, 2009
     

8:30–9:00 

 Continental breakfast  
     
9:00–9:30
 Session 3:  When should sequencing be done to follow up GWAS?
For what situations will the 1000 Genomes data be sufficient for follow-up sequencing of a GWAS?  What range of MAF (for SNPs, structural variants, and haplotypes) will this resource provide?  When will additional sequencing in GWA studies be needed? 
 

Moderators:

 David Altshuler and Lisa Brooks
 

When will 1000 Genomes data suffice?

 Gonçalo Abecasis
   When will sequencing be needed? Gilles Thomas
     
9:30–11:00

Session 4:  How should sequencing be done to follow up GWAS?
What design will be needed?  What information will different designs provide (use both cases and controls; choose samples based on haplotypes, GWAS signals, phenotype extremes; examine exons or entire GWAS hit regions).  What are region boundaries, and how many samples?  How can differences in LD and variation among populations be used?  Will these data allow association of rare variants with phenotypes?
 

Moderators:

 Gonçalo Abecasis and Thomas Lehner
 

Study design

David Altshuler
  Study design    Debbie Nickerson
  Study design Mark McCarthy
  Rare variants David Goldstein
     
11:00–11:15
 Break  
     
11:15–12:15

Session 5:  How should genotyping be done as follow-up to GWAS sequencing? 
How should regions be chosen for dense genotyping (number of new variants found by sequencing, significance of signals)?  How should boundaries of regions be chosen (LD, genetic mapping)?  How should SNPs and structural variants be chosen (type all, choose based on LD and haplotype data, include functional data)?  How many samples should be genotyped, and how should they be chosen (entire GWAS samples or subsets based on phenotypes or exposures, range of populations)? 

 

Moderators:

 Debbie Nickerson and Teri Manolio
  Which regions should be genotyped? David Hunter
  Which variants should be genotyped? Shaun Purcell
  Which samples should be genotyped? John Witte
  Which studies should be genotyped? Chris Haiman
     
12:15–1:00
 Lunch  
     
1:00–1:45

Session 6:  Assessing success: 
What measures will tell us whether we were successful in mapping GWAS signals?  How will we decide whether we should do more?  Is there a limit to nominating variants for laboratory analyses?  How can we maximize the value for experimental functional studies?
 

Moderators:

 Michael Boehnke and Lisa Brooks
  Benchmarks    Gonçalo Abecasis
  Measures of success                          Chris Carlson
  Maximizing the value for functional studies Mila Prokunina-Olsson
     
1:45–3:00
  Conclusion:  What did we learn and what next?
  Moderators:   Stephen Chanock and Lisa Brooks

 

Up to Top

This page last updated: August 7, 2009