NHGRI logo

Genomic Medicine XII: Genomics and Risk Prediction

Event Details

On May 6-7, 2019, the National Human Genome Research Institute (NHGRI) sponsored its 12th Genomic Medicine meeting, Genomic Medicine XII: Genomics and Risk Prediction.

The objectives of the meeting were:

  • Review the state of science of polygenic risk scores and how it can be improved
  • Examine other information sources that should be integrated with genetic variant information in predicting risk
  • Identify research directions in development and implementation of genomic risk prediction


Executive Summary

Meeting Summary


Sheraton Silver Spring
Silver Spring, Maryland


  • Monday, May 6, 2019

  • 8:30 a.m. Welcome and Introductions
    Teri Manolio, Dan Roden

    Video (Roden) - Video (Manolio) - Slides
  • Session 1: Risk prediction with and without genomics
    (Moderator: Teri Manolio)
  • 8:50 a.m. Risk prediction in the Framingham Heart Study
    Adrienne Cupples

    Video - Slides
  • 9:20 a.m. Development and application of polygenic risk scores (PRS) 
    Eric Boerwinkle

    Video - Slides
  • 9:50 a.m. Risk prediction in non-European Ancestry (EA) populations 
    Olufunmilayo Olopade

    Video - Slides
  • 10:20 a.m. Break
  • 10:40 a.m. Current state of risk prediction literature
    Genevive Wojcik

    Video - Slides
  • 11:00 a.m. Discussion 

  • Session 2: Using informatics and electronic health record (EHR) data in risk prediction
    (includes lunch)
    (Moderator: Dan Roden)
  • 11:50 a.m. Phenotype risk scores (PheRS) for risk prediction 
    Josh Denny

    Video - Slides
  • 12:10 p.m. Clinical machine learning for risk prediction
    Mark Craven

    Video - Slides
  • 12:30 p.m. Lunch
  • 1:00 p.m. Predicting risk from multiple Observational Health Data Sciences (OHDSI) databases
    George Hripsak

    Video - Slides
  • 1:20 p.m. Family history risk predictions from the electronic medical records (EMR)    
    Nicholas Tatonetti

    Video - Slides
  • 1:40 p.m. Discussion

  • Session 3: Choosing the best models - lessons from diverse complex diseases
    (Moderator: Mary Relling)
  • 2:20 p.m. Breast Cancer
    Montserrat García-Closas

    Video - Slides
  • 2:40 p.m. Schizophrenia
    Naomi Wray

    Video - Slides
  • 3:00 p.m. Atrial fibrillation
    Patrick Ellinor

    Video - Slides
  • 3:20 p.m. Genomic risk through the lifespan
    Amit Khera

    Video - Slides
  • 3:40 p.m. Discussion 

  • 4:20 p.m. Break
  • 4:40 p.m. Debate: Genomic information is essential to clinical assessment of complex disease risk (Moderator: Howard McLeod)
    Pro: Stephen Chanock
    Con: Isaac Kohane

    Video - Slides (Chanock)
    Video - Slides (Kohane)
  • 5:30 p.m. Discussion

  • 6:00 p.m. Adjourn


  • Tuesday, May 7, 2019

  • Session 4: Other 'omic data
    (Moderator: Pat Deverka)
  • 8:30 a.m. Metabolomics 
    Robert Gerszten

  • 8:50 a.m. Expression Data
    Nancy Cox

    Video - Slides
  • 9:10 a.m. Epigenetics   
    Myriam Fornage

    Video - Slides
  • 9:30 a.m. Environmental exposures
    Peter Kraft

    Video - Slides
  • 9:50 a.m. Discussion

  • 10:30 a.m. Break
  • 11:00 a.m. Panel: Do we need a clinical trial of genomic risk prediction? If so, what should it test, in whom, and with what outcomes? What do we need to know before planning such a trial?
    (Moderator: Howard McLeod)

    Muin Khoury
    Alicia Martin
    George Mensah
    Gina Peloso
    David Valle

    Video - Combined Slides
  • 12:30 p.m. Lunch 
  • 1:00 p.m. Research directions
    (Moderator: Marc Williams)

    Discussion involving all participants to address questions from the sessions to date, such as:

    - What are the critical knowledge gaps? 
    - What impact do age and other factors have on the clinical value of genomic risk prediction?
    - How should other risk factors, such as clinical characteristics, environmental factors, and family history, be integrated with genomic risk prediction? 
    - How and in what contexts should genomic risk prediction be implemented?
    - What thresholds of disease risk or actionability should be considered for implementation?
    - Should genomic risk predictors be assessed for protective effects (i.e., low risk) and should that information be used to modify preventive health recommendations? If so, at what thresholds? 

    Video - Slides

  • 2:00 p.m. Summary and Next Steps
    Teri Manolio, Dan Roden

    Video - Slides
  • 3:00 p.m. Adjourn

Last updated: May 6, 2019