Friday December 6 2019

Session 1 – Predicting risk: from Mendelian to polygenic traits

Shamil Sunyaev - Predicting polygenic risk
Melissa Cline - CAGI Theme: CAGI cancer challenges

Selected talks / flash talks

Lipika Ray - Evolution of CAGI over four rounds: computational challenges in analyzing next generation sequencing data
Natàlia Padilla Sirera - BRCA1- and BRCA2-specific in silico tools for variant interpretation in the CAGI 5 ENIGMA challenge
Andrew Sharo - StrVCTVRE: A supervised learning method to predict the pathogenicity of structural variants
Kymberleigh Pagel - OpenCRAVAT: an open source collaborative platform for the annotation of human genetic variation
Vikas Pejaver - A performance-based approach to establish standards for missense
Yue Cao - Predicting pathogenicity of missense variants with weakly supervised regression
Justin Delano - The impact of missense human variation on post-translational modifications in proteins variant impact prediction tools

Session 2 – Ethical considerations for CAGI

Selected talk: Zhiqiang Hu - Privacy time bombs in omics data: latent risk manifests over time
Barbara Koenig and Malia Fullerton - The CAGI Ethics Forum

Session 3 – Salon talks

Vikas Pejaver - CAGI salon introduction
Kymberleigh Pagel - CAGI challenge design and assessment
Lipida Ray - Broadening impact and participation in CAGI

Saturday December 7 2019

Session 4 – Broader perspectives on Human Genome Variation

Michael Snyder - Big data and health

Resources and policies

Anne O'Donnell-Luria - gnomAD
Marc Greenblatt - ClinGen Sequence Variant Interpretation

Session 5 – Missense variants: promises and limitations

John Moult - CAGI Theme: Bespoke approaches often enhance performance with biophysical methods excelling in a few cases while evolutionary methods have a more consistent performance
Iddo Friedberg - CAGI Theme: Prediction methods have high statistical significance but accuracy is low. However, for an identifiable subset of predictions, accuracy is very high
Olivier Lichtarge - CAGI Theme: Methods tend to correlate with each other more than with experiment

Session 6 – Complex traits and non-coding variants

Sean Mooney - CAGI Theme: Predicting complex traits from exomes is fraught, although there have been improvements in the ability to match genomes to profile
Steve Mount - CAGI Theme: There have been improvements in splicing prediction although this is not yet at the state of missense
Predrag Radivojac - Overcoming (many) challenges in evaluating variant interpretation: a machine learning perspective

Session 7 – Assessment of the annotate all missense CAGI challenge

Nilah Ioannidis - Annotate all missense challenge assessment

Meta-predictor panel – potential, prevarication, and policy
Panelists: Rachel Karchin, Nilah Ioannidis, Sean Mooney
Moderator: Steven Brenner

Sunday December 8 2019

Session 8 – AI and diagnostic variants

Serafim Batzoglou - AI methods for genome interpretation
Constantina Bakolitsa - CAGI Theme: CAGI challenges have led to the identification of causal variants overlooked by clinical labs
Selected talk: Erwin Frise - Benchmarking an artificial intelligence method for fast diagnosis of rare genetic disease

Session 9 – Future perspectives

Future of genome interpretation and vision for CAGI panel
Panelists: Rachel Karchin, Olivier Lichtarge, Vikas Pejaver
Moderator: Steven Brenner